A deep learning based NeuroFusionNet approach for automated brain tumor diagnosis from MRI.

Chapter 12 - A fine-tuned deep transfer learning model in classifying multiclass brain tumors for preclinical MRI image analysis

Brain tumor detection is an active research problem in the field of computer-aided diagnosis in the medical field. While many works used convolutional neural networks (CNN) with transfer learning and addressed this problem with great performance, the interpretability of these transfer learning models was still unclear. In this paper, four different transfer learning settings were tested over three CNN structures including MobileNet, EfficientNet, and ResNet. The first setting is to use a model without transfer learning, the second setting is to use transfer learning keeping all layers learnable; the third setting is to use transfer learning with the first 1/3 layers frozen; the last setting is to use transfer learning with first 2/3 layers frozen. All the pre-trained models were trained on the ImageNet dataset. For each CNN structure and each transfer learning setting, a model was created and trained on the brain Magnetic Resonance Imaging (MRI) dataset. After all the 12 models had been trained, their performance and learned features were compared. Experimental results indicate that the setting with 1/3 layers frozen outperforms other settings, showing the transferability of the first 1/3 layers of models trained on ImageNet to the brain MRI dataset.

Brain tumors are deadly but become deadliest because of delayed and inefficient diagnosis process. Large variations in tumor types also instigate additional complexity. Machine vision brain tumor diagnosis addresses the problem. This research’s objective was to develop a brain tumor classification model based on machine vision techniques using brain Magnetic Resonance Imaging (MRI). For this purpose, a novel hybrid-brain-tumor-classification (HBTC) framework was designed and evaluated for the classification of cystic (cyst), glioma, meningioma (menin), and metastatic (meta) brain tumors. The proposed framework lessens the inherent complexities and boosts performance of the brain tumor diagnosis process. The brain MRI dataset was input to the HBTC framework, pre-processed, segmented to localize the tumor region. From the segmented dataset Co-occurrence matrix (COM), run-length matrix (RLM), and gradient features were extracted. After the application of hybrid multi-features, the nine most optimized features were selected and input to the framework’s classifiers, namely multilayer perception (MLP), J48, meta bagging (MB), and random tree (RT) to classify cyst, glioma, menin, and meta tumors. Maximum brain tumor classification performance achieved by the HBTC framework was 98.8%. The components and performance of the proposed framework show that it is a novel and robust classification framework.

Magnetic Resonance Imaging (MRI) early brain tumor categorization is crucial for the diagnosis of these conditions. Several varieties of diagnostic imaging modalities to detect brain tumours. MRI scans are a highly favoured option for these types of tasks due to their exceptional image quality. Deep learning, a subset of artificial intelligence, has revolutionized the field of automated medical image recognition. The goal of this project was to create a reliable and effective transfer learning technique-based system for MRI-based brain tumour classification. This article describes the development of a brain tumour diagnosis system using common deep learning architectures. Deep features from brain MRIs are extracted using pre-trained models as, Xception, and CNN. Two publicly available benchmark datasets from the internet were used in the experiment. To enhance training speed, accuracy and precision, the image dataset were cropped, pre-processed and enhanced images from the dataset. This study entails utilizing a brain MRI dataset to train and assess deep transfer learning models. This approach employs multiple performance metrics, including accuracy, sensitivity, precision, specificity, and F1-score, to assess its effectiveness. Based on the Xception architecture and utilizing the ADAM optimizer, our Convolutional Neural Network (CNN) model outperforms the other models in our testing data. On the MRI dataset, the Xception model achieved 99.66% accuracy, 99.68% sensitivity, 99.66% precision, 99.68% specificity, and 99.68% F1-score.

Early classification of brain tumors from magnetic resonance imaging (MRI) plays an important role in the diagnosis of such diseases. There are many diagnostic imaging methods used to identify tumors in the brain. MRI is commonly used for such tasks because of its unmatched image quality. The relevance of artificial intelligence (AI) in the form of deep learning (DL) has revolutionized new methods of automated medical image diagnosis. This study aimed to develop a robust and efficient method based on transfer learning technique for classifying brain tumors using MRI. In this article, the popular deep learning architectures are utilized to develop brain tumor diagnostic system. The pre-trained models such as Xception, NasNet Large, DenseNet121 and InceptionResNetV2 are used to extract the deep features from brain MRI. The experiment was performed using two benchmark datasets that are openly accessible from the web. Images from the dataset were first cropped, preprocessed, and augmented for accurate and fast training. Deep transfer learning models are trained and tested on a brain MRI dataset using three different optimization algorithms (ADAM, SGD, and RMSprop). The performance of the transfer learning models is evaluated using performance metrics such as accuracy, sensitivity, precision, specificity and F1-score. From the experimental results, our proposed CNN model based on the Xception architecture using ADAM optimizer is better than the other three proposed models. The Xception model achieved accuracy, sensitivity, precision specificity, and F1-score values of 99.67%, 99.68%, 99.68%, 99.66%, and 99.68% on the MRI-large dataset, and 91.94%, 96.55%, 87.50%, 87.88%, and 91.80% on the MRI-small dataset, respectively. The proposed method is superior to the existing literature, indicating that it can be used to quickly and accurately classify brain tumors.

In today’s world, a brain tumor is one of the most serious diseases. If it is detected at an advanced stage, it might lead to a very limited survival rate. Therefore, brain tumor classification is crucial for appropriate therapeutic planning to improve patient life quality. This research investigates a deep-feature-trained brain tumor detection and differentiation model using classical/linear machine learning classifiers (MLCs). In this study, transfer learning is used to obtain deep brain magnetic resonance imaging (MRI) scan features from a constructed convolutional neural network (CNN). First, multiple layers (19, 22, and 25) of isolated CNNs are constructed and trained to evaluate the performance. The developed CNN models are then utilized for training the multiple MLCs by extracting deep features via transfer learning. The available brain MRI datasets are employed to validate the proposed approach. The deep features of pre-trained models are also extracted to evaluate and compare their performance with the proposed approach. The proposed CNN deep-feature-trained support vector machine model yielded higher accuracy than other commonly used pre-trained deep-feature MLC training models. The presented approach detects and distinguishes brain tumors with 98% accuracy. It also has a good classification rate (97.2%) for an unknown dataset not used to train the model. Following extensive testing and analysis, the suggested technique might be helpful in assisting doctors in diagnosing brain tumors.

Cancer is considered one of the most aggressive and destructive diseases that shortens the average lives of patients. Misdiagnosed brain tumours lead to false medical intervention, which reduces patients' chance of survival. Accurate early medical diagnoses of brain tumour are an essential point for starting treatment plans that improve the survival of patients with brain tumours. Computer-aided diagnostic systems have provided consecutive successes for helping medical doctors make accurate diagnoses and have conducted positive strides in the field of deep and machine learning. Deep convolutional layers extract strong distinguishing features from the regions of interest compared with those extracted using traditional methods. In this study, different experiments are performed for brain tumour diagnosis by combining deep learning and traditional machine learning techniques. AlexNet and ResNet-18 are used with the support vector machine (SVM) algorithm for brain tumour classification and diagnosis. Brain tumour magnetic resonance imaging (MRI) images are enhanced using the average filter technique. Then, deep learning techniques are applied to extract robust and important deep features via deep convolutional layers. The process of combining deep and machine learning techniques starts, where features are extracted using deep learning techniques, namely, AlexNet and ResNet-18. These features are then classified using SoftMax and SVM. The MRI dataset contains 3,060 images divided into four classes, which are three tumours and one normal. All systems have achieved superior results. Specifically, the AlexNet+SVM hybrid technique exhibits the best performance, with 95.10% accuracy, 95.25% sensitivity, and 98.50% specificity.

The role of the pediatrician is crucial in both the diagnosis and management of pediatric brain tumors, the most common solid tumor of childhood. Awareness of the presenting signs and symptoms of brain tumors can lead to timely diagnosis, and understanding the late effects of brain tumor treatment improves long-term management of childhood brain tumor survivors.After completing this article, readers should be able to: Recognize the presenting symptoms and physical examination findings suggestive of a childhood brain tumor and how these findings depend on tumor location.Review common brain tumor pathologies affecting children.Understand how molecular genetics plays a role in the diagnosis and treatment of childhood brain tumors.Recognize the late affects associated with the treatment of childhood brain tumors.Brain tumors are the most common solid malignancy in children and represent the leading cause of pediatric cancer-related deaths. Five thousand new brain tumors are diagnosed yearly in the United States in children ages 0 to 19 years, with an incidence of approximately 6 per 100,000 children. (1) Childhood brain tumors, more than half of which are malignant, vary in terms of biology, prognosis and treatment. Presenting signs and symptoms depend on tumor location, growth rate, and presence of obstructive hydrocephalus. Making the initial diagnosis of a brain tumor can be difficult because early symptoms, such as headaches or vomiting, are nonspecific to brain tumors and more frequently are associated with other etiologies, leading to delays in diagnosis. The pediatrician plays a crucial role in the timely diagnosis of patients with brain tumors as well as recognizing late effects resulting from tumor therapies. This review summarizes the presenting features on history and physical examination, tumor classification of common tumor types, genetic brain tumor predisposition syndromes, general management strategy, and late effects of therapy.Signs and symptoms of a pediatric brain tumor can be nonspecific, insidious, intermittent, and dependent on location within the central nervous system (CNS) and the anatomical pathways affected. Although headache is the most common presenting complaint overall, it is present in only approximately one-third of the children presenting with brain tumors, and, in the absence of other symptoms or physical examination findings, is not in itself predictive of a brain tumor. Elevated intracranial pressure (ICP) is present in approximately half of all children with brain tumors. In addition to headache, it can cause nausea/vomiting, abnormalities of gait and coordination, and papilledema. Vital sign abnormalities associated with increased ICP, known as the Cushing triad (bradycardia, hypertension, abnormal respirations), are late signs of acutely increased ICP but can be absent in those with chronically elevated ICP. In young children with an open fontanelle, macrocephaly, especially when progressive, can be suggestive of hydrocephalus and a potential mass-occupying lesion. (2)Presenting symptoms depend on tumor location (Fig 1), and certain constellations of symptoms can point to specific lesion locations. Table 1 lists commonly overlooked signs and symptoms that can lead to a delayed diagnosis. Wilne et al analyzed presenting features of more than 4,000 childhood brain tumors and found that for posterior fossa tumors, three-quarters presented with nausea and vomiting, two-thirds with headache, three-fifths with abnormal gait and coordination, and one-third with papilledema. (2) In contrast, headache, nausea, and vomiting were rare in patients presenting with supratentorial tumors. Instead, seizures were present in one-third of patients, along with focal neurologic deficits such as weakness or sensory deficits on the contralateral side if there is involvement of the cortical motor or sensory regions, respectively. (2) In cases of brainstem tumors, children can present with crossed findings of ipsilateral facial weakness and contralateral hemiparesis. More than 75% of patients with brainstem tumors present with abnormal gait and coordination, whereas cranial nerve (CN) palsies are present in more than half. Headache, however, is not common in patients with brainstem tumors and is present in less than one-quarter at the time of diagnosis. Thalamic tumors can cause coordination and motor difficulties or hemiplegia. (2)Patients with pituitary tumors or optic pathway tumors often present with visual deficits. It is not uncommon for even severe visual deficits in children to go unrecognized by the patient, parents, or pediatrician. (3) Because patients with neurofibromatosis (NF) type 1 are at increased risk for optic pathway glioma, they should have yearly ophthalmology evaluations. Children with pituitary or hypothalamic tumors often present with endocrine abnormalities, such as failure to thrive, excessive thirst, or central obesity.Children with spinal cord tumors most commonly present with back pain, present at diagnosis in approximately two-thirds of cases. Spinal cord tumors may occur in extradural, intramedullary, and extramedullary intradural locations. Although some children may present with scoliosis, most will not. Spinal cord compression causes signs such as gait and coordination abnormalities, focal weakness, or bowel and bladder dysfunction. (2)A comprehensive neurologic examination (summarized in Table 2) is crucial to identify abnormalities that might be suggestive of a CNS tumor. A normal neurologic examination does not exclude the diagnosis of a brain or spinal cord tumor and must be correlated with symptoms.Patients with acute hydrocephalus can display dramatic changes in their mental status, with increased sleepiness, decreased energy, and decreased responsiveness. However, those with chronic hydrocephalus might show only subtle signs, such as slowly declining school performance.A fundoscopic examination of the optic nerve, CN II, is crucial to assess for papilledema and optic nerve pallor, which can reveal information about hydrocephalus or tumors along the optic pathways. A fundoscopic examination can be difficult in young or uncooperative children, warranting referral to ophthalmology for a dilated examination. Vision should be assessed by confrontation in the 4 quadrants of each eye because different patterns of visual field deficits will suggest varying tumor locations. In younger children, assessment of visual fields can be performed using a colorful object for central fixation and introducing a second object in the periphery and watching for the eyes to track to that object.Eye movements are controlled by CNs III, IV, and VI. The nuclei of CNs III and IV are located in the midbrain, whereas the nucleus of CN VI is in the pons, and brainstem tumors can lead to abnormalities of extraocular movements. Large pineal tumors can cause Parinaud syndrome, characterized by upgaze palsy, convergence-retraction nystagmus, and poorly reactive pupils due to compression of the rostral midbrain. Nystagmus can also be seen in patients with cerebellar tumors or optic pathway tumors.CN V, the trigeminal nerve, has 3 divisions that give sensation to the face. The trigeminal nucleus is located in the pons, as is the nucleus of CN VII (the facial nerve), which controls facial movement. Facial asymmetry or decreased facial sensation should raise concern for a mass in this region. Hearing in each ear should be assessed to look for CN VIII dysfunction.The lower CNs (CNs IX, X, XII) exit from the medulla and are involved in phonation, swallowing, and tongue movement. Palatal asymmetry, change in voice quality, or unilateral glossal atrophy raises suspicion for a medullary lesion. CN XI, the accessory nerve, has the most distal nucleus, also in the medulla, and innervates the trapezius and sternocleidomastoid musculature.Motor function, sensation, and reflexes should be assessed with special attention to comparison with the contralateral findings. Asymmetry can indicate a lesion affecting corticospinal tracts (motor), spinothalamic tracts (temperature, pain, light touch), or dorsal columns (proprioception, vibratory sense). Asymmetrical hyporeflexia can indicate lower motor neuron injury, whereas hyperreflexia and the presence of a Babinski reflex are indicative of upper motor neuron dysfunction. In acute upper motor neuron injury, reflexes may be absent.Patients with cerebellar tumors can present with a wide-based ataxic gait and difficulty with tandem gait. A hemiparetic gait can suggest a tumor involving cortical motor areas, the thalamus, or the brain stem. Patients with cerebellar or brainstem tumors may exhibit abnormal coordination, elicited by testing rapid alternating movements, finger to nose testing, or finger (pointer to thumb) and toe tapping (on the floor) or asking a child to mirror the examiner's finger as the examiner moves the finger laterally and/or vertically.Although not technically part of the neurologic examination, a skin examination is important to assess for dermatologic manifestations of underlying tumor predispositions such as NF type 1 (predisposed to low-grade gliomas [LGGs], especially in optic pathways), NF type 2 (predisposed to acoustic schwannomas and meningiomas), tuberous sclerosis complex (predisposed to subependymal giant cell tumors), or, more rarely, constitutional mismatch repair deficiency syndrome. Patients with constitutional mismatch repair deficiency syndrome have a genetic defect in genes responsible for repairing a specific type of DNA damage known as mismatch repair. Abnormalities in these genes (MLH1, MSH2, MSH5, PMS2) make it more difficult for the body to repair normally occurring DNA damage, leading to mutations and predisposing these patients to many types of cancers at an early age, including brain tumors, most commonly high-grade gliomas (HGGs). (4)The child with a suspected brain tumor might require urgent interventions. Those with unstable vital signs, altered mental status, or concern for increased ICP warrant expedited evaluation, best managed initially in the emergency department. Although magnetic resonance imaging (MRI) with and without contrast is the gold standard imaging technique for optimal visualization for brain tumors and is often needed for neurosurgical planning, in the unstable child, a computed tomographic (CT) scan may be the best initial imaging choice. CT scans can provide information regarding acute hydrocephalus, impending herniation, or acute hemorrhage, all of which represent neurosurgical emergencies. They can also show the anatomical location of a mass, lesion size, presence of hydrocephalus, and whether the mass is compressing other brain structures, thereby helping to triage and plan a timeline for MRI, surgery, or other sedated procedures. When choosing the optimal initial imaging study for a young child who would require anesthesia to complete an MRI, the relative risks of anesthesia compared with the risk of exposure to ionizing radiation from a CT scan, which could be completed without sedation, must be weighed while taking into account the degree of suspicion for an abnormality and individual risk factors specific to that patient. (5)MRI with and without contrast is generally the preferred imaging modality for diagnosis and follow-up of brain tumors. MRI allows for more detailed characterization of the tumor itself and the surrounding anatomy, with more specialized sequences for visualization of edema, relationship to CNs, blood vessels, and perfusion. Furthermore, MRI does not expose children to ionizing radiation so is preferred over CT for repeated studies, as would be needed to follow a brain tumor. Most patients with a brain tumor require a spinal MRI to evaluate for evidence of leptomeningeal disease.When a diagnosis of a brain tumor is made based on imaging, in the absence of a neurosurgical emergency, patients should be managed in concert with neuro-oncology teams preoperatively. Early neuro-oncology consultation allows for additional baseline neurologic examination, can help inform surgical planning based on the working differential diagnosis and postoperative treatment options, and facilitates an opportunity for clinical trial enrollment where presurgical consent may be required.The care of the pediatric neuro-oncology patient requires a multidisciplinary team–based approach. In addition to an excellent primary care pediatrician, this team includes neuro-oncology, neuro-surgery, neurology, neuro-radiology, radiation oncology, genetics, endocrinology, ophthalmology, audiology, neuropsychology, physical medicine and rehabilitation, palliative care, and social work.Upfront treatment of pediatric brain tumors generally includes surgery, radiotherapy, chemotherapy, or a combination of these modalities. For most tumor types, maximal safe surgical resection is pursued to obtain diagnosis and as the first step in definitive treatment. Some notable exceptions to this include tumors in eloquent locations where resection would result in significant morbidity or mortality. These locations include the brain stem, optic pathways, thalamus, internal capsule, sensory and motor cortices, visual cortex, or Broca and Wernicke areas, which are important for receptive and expressive language. In some cases, a small needle biopsy of these areas can be performed to obtain tissue for diagnostic purposes. For germ cell tumors, tumor markers can be diagnostic, obviating the need for upfront surgery. Some patients with low-grade–appearing lesions are followed with observation alone.Although some low-grade tumors can be treated with resection only, many low-grade and most high-grade tumors require additional postsurgical treatment. The standard of care for postsurgical management of pediatric brain tumors is constantly evolving based on emerging preclinical and clinical data. In many cases, enrollment in an open clinical trial is considered the standard of care. There are a variety of clinical trial consortia and cooperative groups with open protocols focused on pediatric brain tumors. A complete list of open clinical trials can be found on clinicaltrials.gov.There are more than 30 unique pathologies of CNS tumors in children. MRI characteristics of some common childhood brain tumors are shown in Fig 2. The advent of molecular genetics has enhanced our understanding of the biologic behavior of brain tumors, has changed tumor classification systems, and has had treatment implications.Medulloblastoma is the most common malignant brain tumor in children and is of embryonal origin. It generally presents as a posterior fossa mass and, due to its location, is often associated with obstructive hydrocephalus. Staging includes an MRI of the spine and a lumbar puncture looking for malignant cells in the cerebrospinal fluid (CSF). Histologically it is classified as classic, large cell anaplastic, or nodular desmoplastic. Overall, medulloblastoma has 5-year overall survival (OS) of approximately 70%. (6)Treatment depends on age at presentation, extent of resection, and presence of metastatic disease. Recent trials are accounting for molecular subtype in treatment decisions. Generally, treatment involves maximal tumor resection, craniospinal radiotherapy, and chemotherapy. Young patients undergo high-dose chemotherapy with autologous stem cell rescue to avoid or delay irradiation.Medulloblastoma has been classified into 4 principle molecular subgroups: WNT (wingless), SHH (sonic hedgehog), group 3, and group 4 (Table 3). (7) WNT-driven medulloblastomas are rarely metastatic and have the best overall prognosis, with greater than 90% OS. Current clinical trials are focused on reducing therapy in this subtype. SHH-driven tumors have a bimodal distribution presenting most commonly in infants or adolescents and young adults. They have an intermediate prognosis, although association with p53 mutations portends a poor prognosis. (9) Group 3 and group 4 tumors are known as non-WNT, non-SHH medulloblastoma subtypes. Although immunohistochemical studies can differentiate WNT and SHH medulloblastoma from the non-WNT and non-SHH medulloblastoma subtypes, other molecular methods, such as methylation studies, are needed to distinguish group 3 from group 4 tumors. Group 3 tumors can present in very young children, often have MYC amplification, are commonly metastatic at presentation, and have the poorest outcomes overall of any subgroup. Recent data suggest that group 3 tumors might benefit from intensified chemotherapy concurrent with radiotherapy. Group 4 tumors are the most common subgroup overall, presenting in children and adults and, similar to group 3 tumors, more commonly present in males than in females. (7) Group 4 tumors have an intermediate prognosis.Atypical teratoid rhabdoid tumors (ATRTs) are also embryonal tumors but can present in the posterior fossa or supratentorial region. These tumors have a very poor prognosis, with 3-year OS of approximately 25%. Survival trends improve with older age at diagnosis, with those older than 3 years faring better than younger patients. (10) Histologically, the loss of INI1, encoded by SMARCB1, is pathognomonic. Up to 35% of patients with ATRT have a germline mutation in SMARCB1 (or rarely SMARCA4), which predisposes them to the development of malignant rhabdoid tumors in other locations, most commonly the kidneys. Germline variants are more common in younger patients, and approximately two-thirds are sporadic. (11)Staging includes MRI of the brain and spine and lumbar puncture for CSF cytology. Treatment involves surgical resection, radiotherapy, and chemotherapy, with or without triple tandem autologous stem cell transplant. Recent clinical trial data showed improved survival outcomes compared with historical controls achieved with a regimen including radiotherapy for patients as young as 6 months and 3 cycles of high-dose chemotherapy with autologous stem cell rescue for all patients. (12) A meta-analysis including 130 patients with ATRT saw that survival correlated most strongly when patients were treated with regimens that included high-dose chemotherapy with autologous stem cell rescue. Treatment modalities of radiotherapy and intrathecal chemotherapy also lead to a statistically significant improvement in OS in this cohort. (10)ATRT tumors have also been classified based on molecular characteristics into 3 subgroups: ATRT–tyrosine (ATRT-TYR), ATRT–sonic hedgehog (ATRT-SHH), and ATRT–myelocytomatosis oncogene (ATRT-MYC), but further research is needed to delineate the prognostic and clinical implications of these subgroups. (13)Ependymoma represents the third most common brain tumor in children and arises from the ependymal cells lining the ventricles or the central canal of the spinal cord. Two-thirds of ependymomas present in the posterior fossa, with the remainder in the supratentorial region or spinal cord. For pediatric ependymoma as a whole, OS at 10 years is approximately 64%, but cases achieving gross total resection followed by radiotherapy fare significantly better. Molecular subtype and gain of chromosome 1q has important prognostic implications as well. (14)Ependymoma is treated with maximal surgical resection followed by focal radiotherapy, except for spinal disease, in which gross total resection without adjuvant radiotherapy can be curative. The role of chemotherapy in ependymoma remains under clinical investigation. Studies have also explored the use of postoperative chemotherapy to delay or omit radiotherapy in patients younger than 3 years, but outcomes were inferior to regimens involving radiotherapy for children older than 12 months. (15)Ependymoma has been divided into 9 molecular subgroups, with 3 subgroups for each anatomical location: spinal, supratentorial, and posterior fossa. Only 6 of the molecular subtypes generally affect children. Pediatric ependymoma of the spine is divided into the subtype and the subtype spinal subtypes have a prognosis. In the posterior fossa, patients with have a prognosis than those with and in the supratentorial those with have OS than those with and are associated with OS less than and survival of approximately are a group of tumors that including tumors the most tumors as and tumors When represent the most common brain tumor in children and can present in many anatomical locations. are less to to other of the CNS than their malignant and in some cases gross total resection can be curative. However, resection is not in certain anatomical locations, such as in the brain stem or with optic pathway common in patients with NF type have a prognosis, with OS of and survival of in a study of with follow-up of when therapy is needed for the regimen of chemotherapy with or and although other chemotherapy regimens have as well. is not in the upfront management of due to for late of the molecular of has that most are by in the most commonly variants and NF type 1 have shown pediatric and have shown in tumors. contrast to pediatric have a prognosis. include high-grade tumors brainstem tumors and of pediatric is tumors may be to surgical is followed by radiotherapy and chemotherapy for these tumors, as a Group study showed improved survival when chemotherapy to radiotherapy compared with radiotherapy specific chemotherapy regimen has as a standard of care for upfront pediatric In contrast, for tumors such as glioma, chemotherapy to radiotherapy has not been shown to survival the to OS and the OS achieved with radiotherapy open based and clinical trials are patients, to improve outcomes for these patients. studies in pediatric that the of pediatric from that of mutations and in tumors, and in tumors, the of in pediatric and a poor prognosis. are from in older children, with significantly improved are more common in children younger than 1 and under are germ cell tumors represent approximately of pediatric brain tumors and are as and germ cell tumors They most commonly in the pineal region but can also present in the thalamus, or or or both or which can be in blood and/or can cause of in the CSF but not In some cases, diagnosis can be made based on CSF and tumor whereas biopsy is when tumor markers are have a better overall prognosis, with OS greater than 90% compared with to for are commonly treated with 4 cycles of chemotherapy followed by radiotherapy to the tumor and whereas are generally treated with 6 cycles of chemotherapy alternating with and craniospinal radiotherapy in many cases, although studies are whether radiotherapy can be in patients with to associated with craniospinal radiotherapy. is a tumor from the of the and solid Histologically, they are classified as tumors and are divided into and subtypes. to their location they can and treatment for is some a more neurosurgical in an to avoid radiotherapy, and an initial resection followed by upfront radiotherapy. are germline mutations that children to specific types of childhood brain tumors in the of tumor predisposition of these is important to the primary care who follow these patients In the child who presents with a brain especially in the of other history of tumors, history of tumors at a young age, or dermatologic findings, it is important to further for these predisposition Children with a known history of predisposition might require genetic for the presence of these syndromes, and specific tumor if found to of these Furthermore, the presence of certain underlying may the of therapy for the management of a brain tumor. Table 4 summarizes germline associated with specific brain tumor the treatment of different tumor types each of the commonly treatment modalities their risks and acute risks of include and damage to structures, as well as morbidity dependent on tumor For posterior fossa syndrome affects an to of patients who undergo resection of large posterior fossa tumors. fossa syndrome is characterized by a combination of or significant with and or motor occurring within 2 of cerebellar and symptoms can months to and many are with deficits. Patients with supratentorial tumors are at greater risk for postoperative seizures and are often on Children with tumors are at increased risk for postoperative visual deficits and tumors by or at a tumor to damage is over to a total to the are without mass, that when at a point to a certain they also radiation at a lower on both the and exit side of the have mass, so the radiation is to within the the of at that the that the effects of radiotherapy are due to the radiation in both and radiotherapy, patients can skin which generally over the treatment Patients intracranial radiotherapy often headache or radiotherapy can cause due to the by the body and can the growth of the resulting in loss of in younger radiotherapy is preferred over radiotherapy, for patients who require craniospinal radiotherapy, because it to important such as the and For patients focal radiotherapy, therapy may to important or result in a significantly overall radiation on tumor the radiation field and for a plan a plan can be to evaluate relative of over based on the brain that would a with each to the use of therapy are the of radiotherapy

Accurate and automated classification of brain tumors from magnetic resonance imaging (MRI) scans is essential for improving diagnostic precision and supporting clinical decision-making. This study presents a deep learning-based framework that employs two convolutional neural network architectures a custom-designed CNN and a pretrained ResNet18 model for multi-class classification of brain tumors using the publicly available Kaggle MRI dataset. The dataset was preprocessed through normalization, augmentation, and resizing to ensure consistency and model generalization. Both models were trained and evaluated using an 80:20 data split, and their performance was assessed based on accuracy, precision, recall, and F1-score metrics. Experimental results demonstrate that the ResNet18 model outperforms the baseline CNN, achieving a classification accuracy of 99.7%, precision of 99.5%, and F1-score of 99.6%. These results highlight the effectiveness of transfer learning and residual connections in improving feature representation and convergence speed. These findings underscore the effectiveness of transfer learning for medical image analysis and demonstrate the potential of deep learning"based methods for reliable, automated brain tumor diagnosis. Future research should focus on extending this work to 3D MRI volumes and integrating explainable AI techniques to enhance interpretability and clinical trust. Accurate classification of brain tumors from Magnetic Resonance Imaging (MRI) is crucial for early diagnosis, treatment planning, and improving patient outcomes. However, manual interpretation of MRI scans is time-consuming and susceptible to diagnostic inconsistencies. This study presents a comparative evaluation of a custom Convolutional Neural Network (CNN) and a transfer learning"based ResNet18 model for automated brain tumor classification using the Kaggle Brain Tumor MRI Dataset, which includes four diagnostic categories: glioma, meningioma, pituitary tumor, and no tumor. Both models were trained and validated under identical preprocessing and experimental conditions to ensure fair comparison. Comprehensive preprocessing, including normalization, augmentation, and stratified splitting (70% training, 20% validation, 10% testing), was applied to enhance data uniformity and generalization. The CNN model was trained from scratch, whereas the ResNet18 model is fine-tuned using pretrained ImageNet weights to leverage transfer learning. Performance was evaluated using accuracy, precision, recall, F1-score, and AUC metrics, supplemented by visual diagnostics such as confusion matrices, accuracy"loss curves, and F1-confidence plots. The ResNet18 model achieved superior performance, with a test accuracy of 99.54%, precision of 0.98, recall of 0.99, F1-score of 0.99, and AUC of 0.992, outperforming the custom CNN, which attained 97.84% accuracy, precision of 0.94, recall of 0.95, F1-score of 0.94, and AUC of 0.975. Confusion matrix analysis indicated that both models accurately classified all tumor types, though minor misclassifications were observed between pituitary and no-tumor categories. ResNet18 exhibited faster convergence, smoother loss reduction, and greater robustness to intensity variations due to its residual connections and pretrained feature representations.

Brain tumor diagnosis is challenging due to complex brain anatomy and tumor variability across imaging views. Traditional methods are manual and error-prone, making deep learning, particularly ResNetbased Convolutional Neural Network (CNN), essential for improving accuracy. The research investigates the enhancement of brain tumor classification using Magnetic Resonance Imaging (MRI) images through a novel modification of the ResNet50 model. It specifically addresses data imbalance challenges in medical image analysis. By proposing a targeted approach to partial data augmentation, the researchers aim to overcome limitations in traditional deep-learning classification methodologies, particularly the performance bottlenecks encountered in differentiating complex brain tumor subtypes. The research uses MRI dataset containing 5,249 labeled images (glioma, meningioma, pituitary, no tumor) across axial, coronal, and sagittal planes, highlighting class and view-based imbalances addressed through targeted augmentation. The research employs transfer learning to analyze three scenarios: non-augmented, partially augmented, and rounding-down data. Results reveal that the partially augmented scenario achieves the highest classification accuracy at 85%, significantly surpassing the non-augmented scenario, which peaks at 79%. In contrast, the rounding-down scenario yields only 60.16% accuracy during validation, highlighting the negative impact of drastically reducing data quantities. The unique contribution lies in demonstrating how strategic partial augmentation can enhance pattern recognition and mitigate overfitting risks, particularly in medical imaging where precise differentiation is crucial. The findings highlight the critical role of nuanced data distribution in enhancing model robustness, as evidenced by improved pattern recognition and reduced overfitting risks in the augmented scenario.

Background/Objectives: The brain serves as the central command center for the nervous system in the human body and is made up of nerve cells known as neurons. When these nerve cells grow rapidly and abnormally, it can lead to the development of a brain tumor. Brain tumors are severe conditions that can significantly reduce a person's lifespan. Failure to detect or delayed diagnosis of brain tumors can have fatal consequences. Accurately identifying and classifying brain tumors poses a considerable challenge for medical professionals, especially in terms of diagnosing and treating them using medical imaging analysis. Errors in diagnosing brain tumors can significantly impact a person's life expectancy. Magnetic Resonance Imaging (MRI) is highly effective in early detection, diagnosis, and classification of brain cancers due to its advanced imaging abilities for soft tissues. However, manual examination of brain MRI scans is prone to errors and heavily depends on radiologists' experience and fatigue levels. Swift detection of brain tumors is crucial for ensuring patient safety. Methods: In recent years, computer-aided diagnosis (CAD) systems incorporating deep learning (DL) and machine learning (ML) technologies have gained popularity as they offer precise predictive outcomes based on MRI images using advanced computer vision techniques. This article introduces a novel hybrid CAD approach named ViT-PCA-RF, which integrates Vision Transformer (ViT) and Principal Component Analysis (PCA) with Random Forest (RF) for brain tumor classification, providing a new method in the field. ViT was employed for feature extraction, PCA for feature dimension reduction, and RF for brain tumor classification. The proposed ViT-PCA-RF model helps detect early brain tumors, enabling timely intervention, better patient outcomes, and streamlining the diagnostic process, reducing patient time and costs. Our research trained and tested on the Brain Tumor MRI (BTM) dataset for multi-classification of brain tumors. The BTM dataset was preprocessed using resizing and normalization methods to ensure consistent input. Subsequently, our innovative model was compared against traditional classifiers, showcasing impressive performance metrics. Results: It exhibited outstanding accuracy, specificity, precision, recall, and F1 score with rates of 99%, 99.4%, 98.1%, 98.1%, and 98.1%, respectively. Conclusions: Our innovative classifier's evaluation underlined our model's potential, which leverages ViT, PCA, and RF techniques, showing promise in the precise and effective detection of brain tumors.

Brain tumor diagnosis is a complex task due to the intricate anatomy of the brain and the heterogeneity of tumors. While magnetic resonance imaging (MRI) is commonly used for brain imaging, accurately detecting brain tumors remains challenging. This study aims to enhance brain tumor classification via deep transfer learning architectures using fine-tuned transfer learning, an advanced approach within artificial intelligence. Deep learning methods facilitate the analysis of high-dimensional MRI data, automating the feature extraction process crucial for precise diagnoses. In this research, several transfer learning models, including InceptionResNetV2, VGG19, Xception, and MobileNetV2, were employed to improve the accuracy of tumor detection. The dataset, sourced from Kaggle, contains tumor and non-tumor images. To mitigate class imbalance, image augmentation techniques were applied. The models were pre-trained on extensive datasets and fine-tuned to recognize specific features in MRI brain images, allowing for improved classification of tumor versus non-tumor images. The experimental results show that the Xception model outperformed other architectures, achieving an accuracy of 96.11%. This result underscores its capability in high-precision brain tumor detection. The study concludes that fine-tuned deep transfer learning architectures, particularly Xception, significantly improve the accuracy and efficiency of brain tumor diagnosis. These findings demonstrate the potential of using advanced AI models to support clinical decision making, leading to more reliable diagnoses and improved patient outcomes.

Early detection and treatment in the medical field offer a critical opportunity to survive people. However, the brain has a significant role in human life as it handles most human body activities. Accurate diagnosis of brain tumors dramatically helps speed up the patient's recovery and the cost of treatment. Magnetic resonance imaging (MRI) is a commonly used technique due to the massive progress of artificial intelligence in medicine, machine learning, and recently, deep learning has shown significant results in detecting brain tumors. This review paper is a comprehensive article suitable as a starting point for researchers to demonstrate essential aspects of using deep learning in diagnosing brain tumors. More specifically, it has been restricted to only detecting brain tumors (binary classification as normal or tumor) using MRI datasets in 2020 and 2021. In addition, the paper presents the frequently used datasets, convolutional neural network architectures (standard and designed), and transfer learning techniques. The crucial limitations of applying the deep learning approach, including a lack of datasets, overfitting, and vanishing gradient problems, are also discussed. Finally, alternative solutions for these limitations are obtained.

Brain tumor detection is a significant problem in medical diagnostics since early and accurate detection improves patient outcomes. Traditional tumor identification techniques often depend on manual interpretation of medical examination, which can be time-consuming and prone to humanerror. Algorithms based on deep learning have emerged in recent years as a viable way to automateand enhance brain tumor identification using medical imaging data. This paper conveys an extensive look on the use of deep learning for brain tumor identification. A Convolutional NeuralNetwork(CNN) architecture is put forward to reach minimum accuracy of 97% and maximum of 100%, using its abilities to automatically learn hierarchical attributes from medical imagery that involve Magnetic Resonance Imaging(MRI) scans. To learn discriminative features suggestive oftumor presence, the suggested CNN framework is trained an extensive collection of labeled brainMRI images. The findings from experiments show that the proposed deep learning approach works. The trained CNN is quite good at differentiating between tumor and non-tumor regions in brain scans. Furthermore, cross-validation and unbiased evaluation are used to assess the model’scapacity to generalize to data that was previously unavailable. Deep learning in brain tumor identification has the potential to greatly enhance diagnostic accuracy, reduce human error, and speed up decision-making. As deep learning research advances, future studies may look at the amalgamation of multi-modal imaging data, transfer learning, and ensemble techniques in order to boost the robustness and generalizability of brain tumor diagnosis. The proposed deep learning-based brain tumor detection system offers the potential for improving medical professionals’ capacity to properly and instantly diagnose brain tumors, ultimately leading to improvements in patient care and outcomes. Keywords: Brain Tumor detection, Diagnosis, Deep Learning, Convolutional NeuralNetworks, Pooling, MRI Dataset

Accurate identification of brain tumors from magnetic resonance imaging (MRI) plays an important role in clinical diagnosis and treatment planning. This paper presents a deep learning–based method for automated brain tumor classification using a Convolutional Neural Network (CNN). The proposed CNN model is trained from scratch on a publicly available brain MRI dataset containing four classes: glioma, meningioma, pituitary tumor, and no tumor. All images are resized to a uniform resolution and processed through an end-to-end learning framework without applying explicit data augmentation. The network learns relevant spatial features through convolutional and pooling layers, followed by fully connected layers for multi-class classification. Experimental results show that the proposed CNN achieves a test accuracy of about 95%, with balanced class-wise performance reflected by a macro-averaged precision, recall, and F1-score of 96%. These findings indicate that CNN-based models can effectively learn meaningful tumor characteristics from MRI scans and may serve as a useful tool to support computer-aided brain tumor diagnosis.