Brain CT Images Research Articles

Coding traumatic brain injury with the abbreviated injury scale following a standardised radiologic template will improve classification of trauma populations.

Injury coding with the Abbreviated Injury Scale (AIS) is an important element for benchmarking, trauma registries and research. To compare the severity of traumatic brain injury (TBI) coding derived from the AIS with or without the use of a standardised radiologic template. A retrospective two-centre cohort study including patients aged ≥ 18 years with isolated TBI admitted to an intensive care between 2011 and 2016 was conducted. TBI was re-coded to conform the AIS by coders, and CT-brain imaging was reassessed by a neuro-radiologist following a standardised radiologic template from which AIS codes were derived. A total of 560 patients were included (median age 57, 37% female). The percentage of MAIS ≥ 4 and major trauma was higher when AIS coding for TBI was derived from a standardised radiologic template vs. coding without (n = 456 (81.4%) and n = 374 (66.8%), p < 0.001; n = 441 (78.8%) and n = 352 (62.9%), p < 0.001, respectively). There was an inter-centre difference in the proportion of MAIS ≥ 4 re-coded without a standardised radiologic template (n = 212 (68.2%) and n = 140 (56.2%), p = 0.004), and no difference when re-coded with the template (n = 251 (80.7%) and n = 190 (76.3%), p = 0.206). Coding TBI with AIS based on a standardised radiologic template results in fewer missed AIS head codes, more detailed AIS head codes, and more patients classified as 'major trauma'. Question Radiologic reports are an important source for injury coding with the abbreviated injury scale (AIS) and are often not sufficiently specific. Findings An AIS-based standardised radiologic template for reporting resulted in more detailed AIS head codes and more patients classified as major trauma. Clinical relevance Injury coding with the AIS based on a standardised radiologic template will improve exchanging medical information in the acute health care setting and classification of trauma populations.

Assessment of a New CT Detector and Filtration Technology: Part 2-Image Quality in Phantoms, Cadavers, and Patients.

The purpose of this work was to evaluate the image quality of a commercial CT scanner equipped with a novel detector and filtration technology called PureVision Optics (PVO). CT number, noise, contrast-to-noise ratio (CNR), modulation transfer function (MTF), and noise power spectrum (NPS) were assessed using the ACR CT Accreditation phantom scanned with various acquisitions at 80 kV, 100 kV, 120 kV, and 135 kV, each with multiple CTDIvol values of 20 mGy, 40 mGy, and 65 mGy. Artifacts were evaluated in an anthropomorphic head phantom, a cadaver head, and in patient studies. Two neuroradiologists assessed image quality features in various patients who were examined with unenhanced brain CT on both scanners. Compared with the conventional scanner, for the same CTDIvol, the PVO scanner produced 20.3% less image noise (P < 0.001), 18.9% higher CNR (P < 0.01), and 24.6% higher spatial resolution (P < 0.001). Streak artifacts were less severe with the PVO scanner for the phantom, cadaver, and patient scans (P < 0.05). Radiologists scored the PVO scanner as significantly better for visualization of the cerebrospinal fluid space over the cerebral sulci in high convexity, image noise in gray and white matter, and artifacts in the posterior fossa. They also significantly preferred the PVO scanner for visualization of the border between brain gray and white matter, cerebrospinal fluid space around the mesencephalon, and overall diagnostic acceptability. For matched CTDIvol values, the scanner equipped with PVO technology produced better objective and subjective image quality metrics in brain CT imaging compared with a conventional CT scanner without PVO. In clinical settings, PVO may allow for lower doses while enhancing imaging through dense areas, improving visualization of subtle details, and offering more effective options for examining obese patients.This research received financial support from Canon Medical Systems USA. The study design and data were fully controlled by the coauthors, of which none are employees or consultants of Canon Medical Systems USA.

Arachnoid Cyst and Autism Spectrum Disorder: A Case Report

Arachnoid cysts (ACs) are cerebrospinal fluid accumulations in the central nervous system, surrounded by thin membranes contiguous with the surrounding normal arachnoid. Although most patients are asymptomatic, some present with various symptoms that sometimes require surgical intervention. The association between arachnoid cysts and autism has been reported in some studies. We present here the case of a 16-year-old female patient, monitored and hospitalized in the day hospital of the child psychiatry department for autism spectrum disorder, who was found to have an arachnoid cyst on brain CT imaging.

Automatic segmentation and volumetric analysis of intracranial hemorrhages in brain CT images

Automatic segmentation and volumetric analysis of intracranial hemorrhages in brain CT images

Brain CT image classification based on mask RCNN and attention mechanism.

Along with the computer application technology progress, machine learning, and block-chain techniques have been applied comprehensively in various fields. The application of machine learning, and block-chain techniques into medical image retrieval, classification and auxiliary diagnosis has become one of the research hotspots at present. Brain tumor is one of the major diseases threatening human life. The number of deaths caused by these diseases is increasing dramatically every year in the world. Aiming at the classification problem of brain CT images in healthcare. We propose a Mask RCNN with attention mechanism method in this research. First, the ResNet-10 is utilized as the backbone model to extract local features of the input brain CT images. In the partial residual module, the standard convolution is substituted by deformable convolution. Then, the spatial attention mechanism and the channel attention mechanism are connected in parallel. The deformable convolution is embedded to the two modules to extract global features. Finally, the loss function is improved to further optimize the precision of target edge segmentation in the Mask RCNN branch. Finally, we make experiments on public brain CT data set, the results show that the proposed image classification fragrance can effectively refine the edge features, increase the degree of separation between target and background, and improve the classification effect.

DDANet: A deep dilated attention network for intracerebral haemorrhage segmentation.

Intracranial haemorrhage (ICH) is an urgent and potentially fatal medical condition caused by brain blood vessel rupture, leading to blood accumulation in the brain tissue. Due to the pressure and damage it causes to brain tissue, ICH results in severe neurological impairment or even death. Recently, deep neural networks have been widely applied to enhance the speed and precision of ICH detection yet they are still challenged by small or subtle hemorrhages. The authors introduce DDANet, a novel haematoma segmentation model for brain CT images. Specifically, a dilated convolution pooling block is introduced in the intermediate layers of the encoder to enhance feature extraction capabilities of middle layers. Additionally, the authors incorporate a self-attention mechanism to capture global semantic information of high-level features to guide the extraction and processing of low-level features, thereby enhancing the model's understanding of the overall structure while maintaining details. DDANetalso integrates residual networks, channel attention, and spatial attention mechanisms for joint optimisation, effectively mitigating the severe class imbalance problem and aiding the training process. Experiments show that DDANet outperforms current methods, achieving the Dice coefficient, Jaccard index, sensitivity, accuracy, and a specificity of 0.712, 0.601, 0.73, 0.997, and 0.998, respectively. The code is available at https://github.com/hpguo1982/DDANet.

Dual-task vision transformer for rapid and accurate intracerebral hemorrhage CT image classification

Intracerebral hemorrhage (ICH) is a severe and sudden medical condition caused by the rupture of blood vessels in the brain, leading to permanent damage to brain tissue and often resulting in functional disabilities or death in patients. Diagnosis and analysis of ICH typically rely on brain CT imaging. Given the urgency of ICH conditions, early treatment is crucial, necessitating rapid analysis of CT images to formulate tailored treatment plans. However, the complexity of ICH CT images and the frequent scarcity of specialist radiologists pose significant challenges. Therefore, we collect a dataset from the real world for ICH and normal classification and three types of ICH image classification based on the hemorrhage location, i.e., Deep, Subcortical, and Lobar. In addition, we propose a neural network structure, dual-task vision transformer (DTViT), for the automated classification and diagnosis of ICH images. The DTViT deploys the encoder from the Vision Transformer (ViT), employing attention mechanisms for feature extraction from CT images. The proposed DTViT framework also incorporates two multilayer perception (MLP)-based decoders to simultaneously identify the presence of ICH and classify the three types of hemorrhage locations. Experimental results demonstrate that DTViT performs well on the real-world test dataset.

Subarachnoid haemorrhage in the emergency department (SHED): a prospective, observational, multicentre cohort study

BackgroundPeople presenting to the ED with acute severe headache often undergo investigation to exclude subarachnoid haemorrhage (SAH). International guidelines propose that brain imaging within 6 hours of headache onset can...

Investigations of Acute Ischemic Stroke Detection Techniques from CT Images Using Machine Learning Approaches

Stroke is a kind of cerebrovascular accident (CVA) that primarily affects individuals over the age of 50, while it can afflict people of all ages. Stroke patients suffer from a chronic condition that leaves them physically unable till their death. To identify the onset of this illness, numerous studies have been carried out. Common signs of a stroke include stiffness, a shift in posture, and tremors in the arms and legs. Functional and emotional mediation are linked to a significant area of the brain. The symptoms of many brain disorders are similar in the early stages because of disruptions to dopamine and other regulating mechanisms. This is a crucial step in how stroke lesions develop. The CAD system is the most efficient method for identifying a stroke. By speeding up the analysis of the required aberration, the CAD system improves the process of illness diagnosis. The goal of the proposed study is to create a compact system that can lessen processing errors, false positive rates, and complexity—three major problems with the current system. The proposed study uses a machine learning model to interpret and classify a CT brain image that shows an Acute Ischaemic Stroke (AIS) lesion. To distinguish between normal and AIS stroke lesions, the pre-processed image is segmented and classed using a Random Forest classifier. Performance criteria like Peak signal ratio, average gradient, accuracy, specificity, sensitivity, and dice index are used to evaluate the experimental output of the proposed work. The suggested model has a straightforward processing structure and the highest efficiency.

Association of the Area Deprivation Index With Dementia Basic Workup and Diagnosis in Central and Western Virginia: A Cross-Sectional Study.

The Area Deprivation Index (ADI) provides a validated and multidimensional metric of areal disadvantage. Our goals were to determine if the ADI influences the likelihood of receiving workup based on published guidelines and an etiologic diagnosis of dementia in Central and Western Virginia. We collected deidentified data from the electronic health record of individuals aged 50-105 years diagnosed with dementia at the University of Virginia (UVA) Medical Center (2016-2021) and at Carillion Clinic (2018-2021). Visit-specific ICD-10 codes were used to classify each dementia diagnosis as "disease-specific" (e.g., Alzheimer disease) or "general" (e.g., unspecified dementia). Following the American Academy of Neurology guidelines, we considered the evaluation performed as "adequate" if patients had vitamin B12, thyroid-stimulating hormone, and brain CT or magnetic resonance imaging within 6 months of the initial diagnosis. Census tract ADI was linked to study participants using the unique census tract identifier derived from the participants' home addresses at the time of diagnosis. Statistical modeling occurred under a Bayesian paradigm implemented using a standard code in R. The study included 13,431 individuals diagnosed with dementia at UVA (n = 7,152) and Carillion Clinic (n = 6,279). Of those, 32.5% and 20.4% received "disease-specific" diagnoses at UVA and Carillion Clinic and 8.2% and 20.4% underwent "adequate" workup, respectively. The adjusted relationship between census tract ADI and the likelihood of a disease-specific diagnosis was U-shaped: Residence in moderately disadvantaged areas was associated with the lowest likelihood of disease-specific diagnosis. Most patients diagnosed with dementia did not receive an adequate evaluation or an etiologic diagnosis. Those living in locations just above the national median ADI levels had the lowest likelihood of receiving an etiologic diagnosis, lower than those in the least and most deprived areas. Renewed awareness efforts among providers are needed to increase compliance with diagnostic guidelines.

Time of Brain Imaging In Patients With An Acute Traumatic Intracranial Haemorrhage

Abstract Background We observed variation in timing of CT brain imaging in patients presenting with a history of trauma. The objective of this audit was to determine adherence to local ED CT Head Criteria in patients with a confirmed traumatic ICH. Methods Retrospective point prevalence audit of patients with a confirmed intracranial haemorrhage over a five-month period using our neurotrauma work-stream data. Exploration of CT timing between Criteria A (within 60 minutes) and Criteria B (within 8 hours). Characteristics reviewed included Gender, Age, Time of presentation to ED, Day of Arrival, GCS on presentation, Mechanism of Injury and Time of CT completion. Results 68 patients over the five-month period had a confirmed ICH: 48 male, 20 female. The median age was 62 years (range 17-97). Falls of less than 2 metres were the most frequent cause of injury (n=44, 65%), with 60% of these occurring in patients aged over 65 years of age. 67% (n=31) of patients meeting Criteria A did not have a CT Brain completed within 60 minutes. 14% (n=3) of patients meeting Criteria B did not have a CT Brain completed within 8 hours. Of patients meeting criteria who did not have a timely CT, 11 patients presented out of hours. Patients aged less than 65 years of age were less likely to meet target CT timing; 71% (n=24). Conclusion This audit highlights that only 33% of patients meeting Criteria A had brain imaging within one hour. Older adults were more likely to meet target CT timing than younger adults. Lack of ED access to radiology out of hours was deemed to be the main barrier to meeting target CT timing. We are working alongside our radiology department to secure no discussion trauma CT scans in ED out of hours. We plan to re-audit following the implementation of this change.

Identification of Bipolar Disorder and Schizophrenia Based on Brain CT and Deep Learning Methods.

With the increasing prevalence of mental illness, accurate clinical diagnosis of mental illness is crucial. Compared with MRI, CT has the advantages of wide application, low price, short scanning time, and high patient cooperation. This study aims to construct a deep learning (DL) model based on CT images to make identification of bipolar disorder (BD) and schizophrenia (SZ). A total of 506 patients (BD = 227, SZ = 279) and 179 healthy controls (HC) was collected from January 2022 to May 2023 at two hospitals, and divided into an internal training set and an internal validation set according to a ratio of 4:1. An additional 65 patients (BD = 35, SZ = 30) and 40 HC were recruited from different hospitals, and served as an external test set. All subjects accepted the conventional brain CT examination. The DenseMD model for identify BD and SZ using multiple instance learning was developed and compared with other classical DL models. The results showed that DenseMD performed excellently with an accuracy of 0.745 in the internal validation set, whereas the accuracy of the ResNet-18, ResNeXt-50, and DenseNet-121model was 0.672, 0.664, and 0.679, respectively. For the external test set, DenseMD again outperformed other models with an accuracy of 0.724; however, the accuracy of the ResNet-18, ResNeXt-50, and DenseNet-121model was 0.657, 0.638, and 0.676, respectively. Therefore, the potential of DL models for identification of BD and SZ based on brain CT images was established, and identification ability of the DenseMD model was better than other classical DL models.

Cross-Modality Image Translation From Brain 18 F-FDG PET/CT Images to Fluid-Attenuated Inversion Recovery Images Using the CypixGAN Framework.

PET/CT and MRI can accurately diagnose dementia but are expensive and inconvenient for patients. Therefore, we aimed to generate synthetic fluid-attenuated inversion recovery (FLAIR) images from 18 F-FDG PET and CT images of the human brain using a generative adversarial network (GAN)-based deep learning framework called the CypixGAN, which combined the CycleGAN framework with the L1 loss function of the pix2pix. Data from 143 patients who underwent PET/CT and MRI were used for training (n = 79), validation (n = 20), and testing (n = 44) the deep learning frameworks. Synthetic FLAIR images were generated using the pix2pix, CycleGAN, and CypixGAN, and white matter hyperintensities (WMHs) were then segmented. The performance of CypixGAN was compared with that of the other frameworks. The CypixGAN outperformed the pix2pix and CycleGAN in generating synthetic FLAIR images with superior visual quality. Peak signal-to-noise ratio and structural similarity index (mean ± standard deviation) estimated using the CypixGAN (20.23 ± 1.31 and 0.80 ± 0.02, respectively) were significantly higher than those estimated using the pix2pix (19.35 ± 1.43 and 0.79 ± 0.02, respectively) and CycleGAN (18.74 ± 1.49 and 0.78 ± 0.02, respectively) ( P < 0.001). WMHs in synthetic FLAIR images generated using the CypixGAN closely resembled those in ground-truth images, as indicated by the low absolute percentage volume differences and high dice similarity coefficients. The CypixGAN generated high-quality FLAIR images owing to the preservation of spatial information despite using unpaired images. This framework may help improve diagnostic performance and cost-effectiveness of PET/CT when MRI scan is unavailable.

Automated ASPECTS Segmentation and Scoring Tool: a Method Tailored for a Colombian TelestrokeNetwork.

To evaluate our two non-machine learning (non-ML)-based algorithmic approaches for detecting early ischemic infarcts on brain CT images of patients with acute ischemic stroke symptoms, tailored to our local population, to be incorporated in our telestroke software. One-hundred and thirteen acute stroke patients, excluding hemorrhagic, subacute, and chronic patients, with accessible brain CT images were divided into calibration and test sets. The gold standard was determined through consensus among three neuroradiologist. Four neuroradiologist independently reported Alberta Stroke Program Early CT Scores (ASPECTSs). ASPECTSs were also obtained using a commercial ML solution (CMLS), and our two methods, namely the Mean Hounsfield Unit (HU) relative difference (RELDIF) and the density distribution equivalence test (DDET), which used statistical analyze the of the HUs of each region and its contralateral side. Automated segmentation was perfect for cortical regions, while minimal adjustment was required for basal ganglia regions. For dichotomized-ASPECTSs (ASPECTS < 6) in the test set, the area under the receiver operating characteristic curve (AUC) was 0.85 for the DDET method, 0.84 for the RELDIF approach, 0.64 for the CMLS, and ranged from 0.71-0.89 for the neuroradiologist. The accuracy was 0.85 for the DDET method, 0.88 for the RELDIF approach, and was ranged from 0.83 - 0.96 for the neuroradiologist. Equivalence at a margin of 5% was documented among the DDET, RELDIF, and gold standard on mean ASPECTSs. Noninferiority tests of the AUC and accuracy of infarct detection revealed similarities between both DDET and RELDIF, and the CMLS, and with at least one neuroradiologist. The alignment of our methods with the evaluations of neuroradiologist and the CMLS indicates the potential of our methods to serve as supportive tools in clinical settings, facilitating prompt and accurate stroke diagnosis, especially in health care settings, such as Colombia, where neuroradiologist are limited.

A ResNet based transfer learning pulmonary medical image disease detection system

Medical imaging represents a significant application of deep learning. In response to the vast domain of medical image analysis and the limitations and inefficiencies of conventional methods for diagnosing brain diseases, this paper explores the application of an enhanced ResNet-50 model in the classification of brain CT images. The goal is to improve the detection and classification accuracy of aneurysms, cancer, and malignant tumors. The study utilized 259 images, undergoing training, validation, and testing processes to verify the model's efficacy. The ResNet-50 model addresses the issue of vanishing gradients in deep networks through residual learning, making it suitable for high-resolution images and small datasets. Results indicate good training performance, though some fluctuations in validation performance, achieving a validation accuracy of 0.960. This research not only enhances diagnostic accuracy for brain diseases but also paves a new path for the further application of deep learning technologies in the medical field.

Prognostic Value of Optic Nerve Sheath Diameters after Acute Ischemic Stroke According to Slice Thickness on Computed Tomography.

The optic nerve sheath diameter (ONSD) can predict intracranial pressure and outcomes in neurological disease, but it remains unclear whether a small ONSD can be accurately measured on routine CT images with a slice thickness of approximately 4-5 mm. We measured the ONSD and ONSD/eyeball transverse diameter (ETD) ratio on routine-slice (4 mm) and thin-slice (0.6-0.75 mm) brain CT images from initial scans of acute ischemic stroke (AIS) patients. ONSD-related variables, National Institutes of Health Stroke Scale (NIHSS) scores, and age were compared between good (modified Rankin Scale [mRS] ≤ 2) and poor (mRS > 2) outcomes at discharge. Among 155 patients, 38 had poor outcomes. The thin-slice ONSD was different between outcome groups (p = 0.047), while the routine-slice ONSD showed no difference. The area under the curve (AUC) values for the ONSD and ONSD/ETD were 0.58 (95% CI, 0.49-0.66) and 0.58 (95% CI, 0.50-0.66) on the routine-slice CT, and 0.60 (95% CI, 0.52-0.68) and 0.62 (95% CI, 0.54-0.69) on the thin-slice CT. The thin-slice ONSD/ETD ratio correlated with initial NIHSS scores (r = 0.225, p = 0.005). After adjusting for NIHSS scores and age, ONSD-related variables were not associated with outcomes, and adding them to a model with NIHSS scores and age did not improve performance (all p-values > 0.05). Although ONSD measurements were not an independent outcome predictor, they correlated with stroke severity, and the thin-slice ONSD provided a slightly better prognostic performance than the routine-slice ONSD.

Ensemble machine learning to predict futile recanalization after mechanical thrombectomy based on non-contrast CT imaging

ObjectivesDespite successful recanalization after Mechanical Thrombectomy (MT), approximately 25 % of patients with Acute Ischemic Stroke (AIS) due to Large Vessel Occlusion (LVO) show unfavorable clinical outcomes, namely Futile Recanalization (FR). We aimed to use a Machine Learning (ML) Non-Contrast brain CT (NCCT) imaging predictive model to identify FR in patients undergoing MT. Materials & methodsBetween July 2022 and December 2022, 70 consecutive patients with LVO undergoing a complete recanalization (eTICI 3) with MT within 8 h from onset at our Centre were analyzed. Two NCCT images per patient of middle cerebral artery vascular territory and patients' clinical characteristics were classified by the presence of ischemic features on 24 h NCCT after MT. Each slice was segmented with “Mazda” software ver.4.6 by placing a Region Of Interest (ROI) on the whole brain by two radiologists in consensus. A total of 381 features were extracted for each slice. The dataset was split into train and test set with a 70:30 ratio. ResultsEleven classification models were trained. An Ensemble Machine Learning (EML) model was obtained by averaging the predictions of models with accuracy on a test set >70 %, with and without patients' clinical characteristics.The EML model combined with clinical data showed an accuracy of 0.76, a sensitivity of 0.88, a specificity of 0.69 with a NPV of 0.90, a PPV of 0.64, with AUC of 0.84. ConclusionNCCT and ML analysis shows promise in predicting FR after complete recanalization following MT in AIS patients. Larger studies are required to confirm these preliminary results.

Optimal image quality and radiation doses with optimal tube voltages/currents for pediatric anthropomorphic phantom brains.

Using pediatric anthropomorphic phantoms (APs), we aimed to determine the scanning tube voltage/current combinations that could achieve optimal image quality and avoid excessive radiation exposure in pediatric patients. A 64-slice scanner was used to scan a standard test phantom to determine the volume CT dose indices (CTDIvol), and three pediatric anthropomorphic phantoms (APs) with highly accurate anatomy and tissue-equivalent materials were studied. These specialized APs represented the average 1-year-old, 5-year-old, and 10-year-old children, respectively. The physical phantoms were constructed with brain tissue-equivalent materials having a density of ρ = 1.07 g/cm3, comprising 22 numbered 2.54-cm-thick sections for the 1-year-old, 26 sections for the 5-year-old, and 32 sections for the 10-year-old. They were scanned to acquire brain CT images and determine the standard deviations (SDs), effective doses (EDs), and contrast-to noise ratios (CNRs). The APs were scanned by 21 combinations of tube voltages/currents (80, 100, or 120 kVp/10, 40, 80, 120, 150, 200, or 250 mA) and rotation time/pitch settings of 1 s/0.984:1. The optimal tube voltage/current combinations yielding optimal image quality were 80 kVp/80 mA for the 1-year-old AP; 80 kVp/120 mA for the 5-year-old AP; and 80 kVp/150 mA for the 10-year-old AP. Because these scanning tube voltages/currents yielded SDs, respectively, of 12.81, 13.09, and 12.26 HU, along with small EDs of 0.31, 0.34, and 0.31 mSv, these parameters and the induced values were expediently defined as optimal. The optimal tube voltages/currents that yielded optimal brain image quality, SDs, CNRs, and EDs herein are novel and essentially important. Clinical translation of these optimal values may allow CT diagnosis with low radiation doses to children's heads.

CT histogram analysis to distinguish between acute intracerebral hemorrhage and cavernous hemangioma

Acute intracerebral hemorrhage (AICH) and cerebral cavernous hemangioma (CCM) are two common cerebral hemorrhage diseases with partially overlapping CT findings and clinical symptoms, making it hard to distinguish between them. The current study used histogram analysis based on CT images to differentiate between CCM and AICH and test its diagnosis performance. This retrospective study included 158 patients with CCM and 137 patients with AICH. The histograms of brain CT plain scan images of both groups were extracted using Python code and included 18 histogram parameters of the lesions. The most effective parameters were selected by univariate logistic regression analysis and Spearman correlation analysis and included in the final multivariate logistic regression model. The sample was randomly divided into the training set and the validation set by 7:3. The ROC curve was constructed to evaluate the discriminant efficiency of the final logistic regression model in distinguishing between AICH and CCM. The univariate analysis identified seven significant histogram parameters with the following final logistic regression model: F=3.731+2.6411×10-9×Energy-1.192×Kurtosis-0.003×Minimum-1.449×Skewness + 2.5002×10-10×Total Energy-1.103×Uniformity+0.009×Variance. The model showed good diagnostic performance in distinguishing between AICH and CCM, with an AUC of 0.876, sensitivity of 70.8%, and specificity of 91.9% in the training set, and an AUC of 0.870, sensitivity of 82.9%, and specificity of 85.1% in the validation set. The histogram analysis of brain CT images can be used as an auxiliary method to distinguish between AICH and CCM effectively.

Classification of recurrent depression using brain CT images through feature fusion

ObjectiveThe objective of the study is to address the issues of high workload and low diagnostic efficiency in clinical medicine by employing computer-aided diagnostic analysis methods. The focus is on the identification of depressed patients using medical image data, particularly brain CT images. MethodsThe study proposes a CT image classification method for identifying depression disorder based on deep learning theory. Key methods include migration learning and feature fusion. Preprocessing and data enhancement techniques are utilized to filter out unnecessary features. An attention module is employed to better extract deep feature information. Additionally, a binary classification focus loss function is applied to address the unbalanced distribution of the dataset. ResultsExperimental results indicate that the proposed method achieves an image classification accuracy of 97.79%. Compared to a single model, there is an increase in accuracy by 2.61% and 1.81%, respectively. This improvement effectively enhances the classification accuracy of brain CT images with depressive disorders. The classification model also demonstrates better generalization performance. ConclusionThe study concludes that the proposed CT image classification method, incorporating migration learning, feature fusion, preprocessing, and attention modules, significantly improves the accuracy and classification speed of identifying depressive disorders in brain CT images. The model's generalization performance is highlighted, providing effective support for the enhancement of medical image classification accuracy.