Radiologists In Detection Research Articles

Artificial Intelligence-based Software for Breast Arterial Calcification Detection on Mammograms.

The performance of a commercially available artificial intelligence (AI)-based software that detects breast arterial calcifications (BACs) on mammograms is presented. This retrospective study was exempt from IRB approval and adhered to the HIPAA regulations. Breast arterial calcification detection using AI was assessed in 253 patients who underwent 314 digital mammography (DM) examinations and 143 patients who underwent 277 digital breast tomosynthesis (DBT) examinations between October 2004 and September 2022. Artificial intelligence performance for binary BAC detection was compared with ground truth (GT) determined by the majority consensus of breast imaging radiologists. Area under the receiver operating curve (AUC), sensitivity, specificity, positive predictive value and negative predictive value (NPV), accuracy, and BAC prevalence rates of the AI algorithm were compared. The case-level AUCs of AI were 0.96 (0.93-0.98) for DM and 0.95 (0.92-0.98) for DBT. Sensitivity, specificity, and accuracy were 87% (79%-93%), 92% (88%-96%), and 91% (87%-94%) for DM and 88% (80%-94%), 90% (84%-94%), and 89% (85%-92%) for DBT. Positive predictive value and NPV were 82% (72%-89%) and 95% (92%-97%) for DM and 84% (76%-90%) and 92% (88%-96%) for DBT, respectively. Results are 95% confidence intervals. Breast arterial calcification prevalence was similar for both AI and GT assessments. Breast AI software for detection of BAC presence on mammograms showed promising performance for both DM and DBT examinations. Artificial intelligence has potential to aid radiologists in detection and reporting of BAC on mammograms, which is a known cardiovascular risk marker specific to women.

Does sevoflurane sedation in pediatric patients lead to "pseudo" leptomeningeal enhancement in the brain on 3 Tesla magnetic resonance imaging?

Prominent leptomeningeal contrast enhancement (LMCE) in the brain is observed in some pediatric patients during sedation for imaging. However, based on clinical history and cerebrospinal fluid analysis, the patients are not acutely ill and do not exhibit meningeal signs. Our study determined whether sevoflurane inhalation in pediatric patients led to this pattern of 'pseudo' LMCE (pLMCE) on 3 Tesla magnetic resonance imaging (MRI). To highlight the significance of pLMCE in pediatric patients undergoing enhanced brain MRI under sedation to avoid misinterpretation in reports. A retrospective cross-sectional evaluation of pediatric patients between 0-8 years of age was conducted. The patients underwent enhanced brain MRI under inhaled sevoflurane. The LMCE grade was determined by two radiologists, and interobserver variability of the grade was calculated using Cohen's kappa. The LMCE grade was correlated with duration of sedation, age and weight using the Spearman rho rank correlation. A total of 63 patients were included. Fourteen (22.2%) cases showed mild LMCE, 48 (76.1%) cases showed moderate LMCE, and 1 case (1.6%) showed severe LMCE. We found substantial agreement between the two radiologists in detection of pLMCE on post-contrast T1 imaging (kappa value = 0.61; P < 0.001). Additionally, we found statistically significant inverse and moderate correlations between patient weight and age. There was no correlation between duration of sedation and pLMCE. pLMCE is relatively common on post-contrast spin echo T1-weighted MRI of pediatric patients sedated by sevoflurane due to their fragile and immature vasculature. It should not be misinterpreted for meningeal pathology. Knowing pertinent clinical history of the child is an essential prerequisite to avoid radiological overcalling and the subsequent burden of additional investigations.

Comparison of Two-Point Compression Ultrasonography Performed by Emergency Medicine Doctors and Radiologists in Detection of Deep Vein Thrombosis

AbstractVenous thromboembolism (VTE) is a serious coagulation disorder that includes deep vein thrombosis (DVT) and pulmonary embolism and is an important cause of hospitalization and death. This study aimed to evaluate the compliance of the emergency doctor and radiologist in diagnosing DVT in the emergency department using the two-point compression ultrasonography (USG) method. Patients and Methods This prospective cross-sectional study was performed between February and July 2022 in the Emergency Medicine Clinic of a tertiary university hospital with patients who were thought to have DVT and had lower extremity venous USG indication. Demographic information of patients, clinical markers used in the Wells score, and USG results of the emergency doctors and radiologists were recorded in the study form. Results A total of 400 patients were included in the study. The mean age of the study patients was 59.8 ± 18.0 years, and 54.4% (n = 217) of the patients were male. There was a significant difference in the incidence of DVT between those with a Wells score of 2 or less and those above 2 (n = 67, 21.8% vs. n = 41, 47.1%; p &lt; 0.001). Regarding interobserver agreement in the evaluation of DVT by emergency medicine doctor and radiologist, kappa values were 0.81 (95% confidence interval [CI]: 0.71–0.91) for the right femoral vein, 0.89 (95% CI: 0.81–0.97) for the left femoral vein. It was found to be 0.81 (95% CI: 0.76–0.86) for all lower extremity vein USGs. Conclusions There is a very good level of agreement between the emergency department and the radiologist in diagnosing DVT with USG.

EP01.04-003 Effectiveness of Cloud-based Computer Aided Quality Control System in Korean National Lung Cancer Screening

Korean national lung cancer screening program (KNLCS) targeting high-risk smoking population using low-dose CT (LDCT) was implemented in 2019. A cloud-based quality control system (CQCS) using computer aided detection program (CAD) was used to assist radiologists in detection, measurement and categorization of lung nodules in LDCT. From 2021, Artificial Intelligence (AI)-based CAD has been launched as a developed version of CQCS. This study aimed to evaluate the effectiveness of CQCS on positive rate and inter-observer variability in KNLCS on national level.

Automated Stanford classification of aortic dissection using a 2-step hierarchical neural network at computed tomography angiography.

This study aimed to evaluate the feasibility of automatic Stanford classification of classic aortic dissection (AD) using a 2-step hierarchical neural network. Between 2015 and 2019, 130 arterial phase series (57 type A, 43 type B, and 30 negative cases) in aortic CTA were collected for the training and validation. A 2-step hierarchical model was built including the first step detecting AD and the second step predicting the probability (0-1) of Stanford types. The model's performance was evaluated with an off-line prospective test in 2020. The sensitivity and specificity for Stanford type A, type B, and no AD (Sens A, B, N and Spec A, B, N, respectively) and Cohen's kappa were reported. Of 298 cases (22 with type A, 29 with type B, and 247 without AD) in the off-line prospective test, the Sens A, Sens B, and Sens N were 95.45% (95% confidence interval [CI], 77.16-99.88%), 79.31% (95% CI, 60.28-92.01%), and 93.52% (95% CI, 89.69-96.25%), respectively. The Spec A, Spec B, and Spec N were 98.55% (95% CI, 96.33-99.60%), 94.05% (95% CI, 90.52-96.56%), and 94.12% (95% CI, 83.76-98.77%), respectively. The classification rate achieved 92.28% (95% CI, 88.64-95.04%). The Cohen's kappa was 0.766 (95% CI, 0.68-0.85; p < 0.001). Stanford classification of classic AD can be determined by a 2-step hierarchical neural network with high sensitivity and specificity of type A and high specificity in type B and no AD. • The Stanford classification for aortic dissection is widely adopted and divides it into Stanford type A and type B based on the ascending thoracic aorta dissected or not. • The 2-step hierarchical neural network for Stanford classification of classic aortic dissection achieved high sensitivity (95.45%) and specificity (98.55%) of type A and high specificity in type B and no aortic dissection (94.05% and 94.12%, respectively) in 298 test cases. • The 2-step hierarchical neural network demonstrated moderate agreement (Cohen's kappa: 0.766, p < 0.001) with cardiovascular radiologists in detection and Stanford classification of classic aortic dissection in 298 test cases.

“ROLE OF COMPUTER AIDED DIAGNOSIS (CAD) IN EVALUATING LUNG LESIONS – AN INSTITUTIONAL EXPERIENCE”

Introduction: Lung opacities are one of most common radiological ndings and can be seen in variety of lung pathologies. They are classied into ground glass opacities, nodules and masses depending on size. CT has been the imaging modality of choice for the detection, characterization and follow-up of lung nodules. Traditionally these tasks are performed by Radiologists , however recently advances like Computer-Aided Diagnosis (CAD) are gaining interest to reduce the load on Radiologists. Aim: To study the role of Computer Aided Diagnosis (CAD) in evaluating lung lesions on Chest Computed Tomography (CT). Materials and Methods:A retrospective analysis conducted in department of Radiology in a Tertiary care Hospital, Hyderabad over a period of 3 months . A total of 100 patients with lung lesions ( ground glass opacities (GGOs), nodules and masses ) who underwent Chest CT were studied. Results: In our study, 18 patients had GGOs, 65 had nodules and 17 had masses. CAD could detect GGOs only in 04 patients, whereas Radiologist detected GGOs in 16 patients. For nodules, we analysed size, number, location &amp; morphology for which CAD showed overall poor performance as compared to Radiologist. For masses, CAD and Radiologist could detect lesions in all 17 patients. Conclusions: In our study, the role of CAD is only a complimentary tool to the Radiologist in detection and characterization of lung lesions and thereby enhancing the workow.

Performance of Deep Learning and Genitourinary Radiologists in Detection of Prostate Cancer Using 3-T Multiparametric Magnetic Resonance Imaging.

Several deep learning-based techniques have been developed for prostate cancer (PCa) detection using multiparametric magnetic resonance imaging (mpMRI), but few of them have been rigorously evaluated relative to radiologists' performance or whole-mount histopathology (WMHP). To compare the performance of a previously proposed deep learning algorithm, FocalNet, and expert radiologists in the detection of PCa on mpMRI with WMHP as the reference. Retrospective, single-center study. A total of 553 patients (development cohort: 427 patients; evaluation cohort: 126 patients) who underwent 3-T mpMRI prior to radical prostatectomy from October 2010 to February 2018. 3-T, T2-weighted imaging and diffusion-weighted imaging. FocalNet was trained on the development cohort to predict PCa locations by detection points, with a confidence value for each point, on the evaluation cohort. Four fellowship-trained genitourinary (GU) radiologists independently evaluated the evaluation cohort to detect suspicious PCa foci, annotate detection point locations, and assign a five-point suspicion score (1: least suspicious, 5: most suspicious) for each annotated detection point. The PCa detection performance of FocalNet and radiologists were evaluated by the lesion detection sensitivity vs. the number of false-positive detections at different thresholds on suspicion scores. Clinically significant lesions: Gleason Group (GG)≥ 2 or pathological size≥ 10 mm. Index lesions: the highest GG and the largest pathological size (secondary). Bootstrap hypothesis test for the detection sensitivity between radiologists and FocalNet. For the overall differential detection sensitivity, FocalNet was 5.1% and 4.7% below the radiologists for clinically significant and index lesions, respectively; however, the differences were not statistically significant (P= 0.413 and P=0.282, respectively). FocalNet achieved slightly lower but not statistically significant PCa detection performance compared with GU radiologists. Compared with radiologists, FocalNet demonstrated similar detection performance for a highly sensitive setting (suspicion score≥ 1) or a highly specific setting (suspicion score=5), while lower performance in between. 3 TECHNICAL EFFICACY STAGE: 2.

Venous Thrombosis and Hypercoagulability in the Abdomen and Pelvis: Causes and Imaging Findings.

Venous thromboembolism (VTE), which includes deep venous thrombosis and pulmonary embolism, is a significant cause of morbidity and mortality. In recent decades, US, CT, and MRI have surpassed catheter-based angiography as the imaging examinations of choice for evaluation of vascular structures and identification of thrombus owing to their ready availability, noninvasive nature, and, in the cases of US and MRI, lack of exposure to ionizing radiation. As a result, VTE and associated complications are commonly identified in day-to-day radiologic practice across a variety of clinical settings. A wide range of hereditary and acquired conditions can increase the risk for development of venous thrombosis, and many patients with these conditions may undergo imaging for unrelated reasons, leading to the incidental detection of VTE or one of the associated complications. Although the development of VTE may be an isolated occurrence, the imaging findings, in conjunction with the clinical history and vascular risk factors, may indicate a predisposing condition or underlying diagnosis. Furthermore, awareness of the many clinical conditions that result in an increased risk of venous thrombosis may aid in detection of thrombus and any concomitant complications. For these reasons, it is important that practicing radiologists be familiar with the multimodality imaging findings of thrombosis, understand the spectrum of diseases that contribute to the development of thrombosis, and recognize the potential complications of hypercoagulable states and venous thrombosis. Online DICOM image stacks and supplemental material are available for this article. ©RSNA, 2020.

Detection of Lung Diseases using Deep Learning

There is an increasing risk on health due to changing environment, climate and lifestyle. India tops the world in deaths due to lung diseases. They were the second highest cause of deaths in India after heart disease in 2017, killing 1 million (958,000) Indians that year. Early diagnosis and treatment of lung diseases is critical to prevent complications including death. Chest X-ray is currently the best available method for diagnosis, playing a crucial role in clinical care. Using Deep Learning to predict lung diseases from Chest Xrays can be a lifesaving factor for an individual suffering from the disease. This is possible as the results can be predicted with a high percentage of accuracy instantly. This paper presents an effective way for expert diagnosis of lung diseases using Deep Learning. It focuses on creating a system for assistance of Radiologists in detection of lung diseases. This will especially benefit rural areas where radiologists aren’t easily available. Our system connects radiology labs with the radiologists who can diagnose faster and better with our model.

Radiologist Identification of Occult Dorsal Wrist Ganglion Cysts on MRI.

Background Occult dorsal ganglion cysts (ODGC) require imaging or an operation for detection. It is currently unknown how often a radiologist identifies an ODGC on magnetic resonance imaging (MRI) and whether surgeon indication for MRI aids the radiologist in detection. Questions/Purposes The aim of the study was to investigate the following questions: how accurately do radiologists identify ODGC on MRIs? What factors may be associated with missed ODGCs? Patients and Methods We retrospectively studied patients who underwent an operation for an ODGC and had a preoperative wrist MRI. Radiology report and surgeon's notes were evaluated to determine whether identification of the ODGC was noted and whether the surgeon indicated that the MRI was to evaluate for an ODGC. MRIs were reviewed, dimensions of cysts measured and volume of cyst calculated. Results Twenty-four patients and 25 MRIs were analyzed. The radiologist identified the ODGC in 19 cases (76%). Fifteen of the 25 MRIs (60%) ordered had ODGC listed in the requisition comment by the ordering surgeon. Thirteen of these 15 (87%) ODGCs were seen by the radiologist. Ten of the 25 MRIs (40%) ordered did not mention ODGC in the requisition. Six of these 10 (60%) ODGCs were seen by the radiologist. The volume of the ODGCs missed by radiologists was smaller (mean, 0.049 cm 3 ) than those the radiologists identified (mean, 0.31 cm 3 ; p = 0.004). Conclusions Radiologists will not always identify the ODGC on an MRI, but they were more likely to if the surgeon was concerned for one. Hand surgeons should report suspicion of an ODGC on MRI requisition and review all imaging independently. Level of Evidence This is a Level III, prognostic study.

Multi-parametric MRI zone-specific diagnostic model performance compared with experienced radiologists for detection of prostate cancer

ObjectivesCompare the performance of zone-specific multi-parametric-MRI (mp-MRI) diagnostic models in prostate cancer detection with experienced radiologists.MethodsA single-centre, IRB approved, prospective STARD compliant 3 T MRI test dataset of 203 patients was generated to test validity and generalisability of previously reported 1.5 T mp-MRI diagnostic models. All patients included within the test dataset underwent 3 T mp-MRI, comprising T2, diffusion-weighted and dynamic contrast-enhanced imaging followed by transperineal template ± targeted index lesion biopsy. Separate diagnostic models (transition zone (TZ) and peripheral zone (PZ)) were applied to respective zones. Sensitivity/specificity and the area under the receiver operating characteristic curve (ROC-AUC) were calculated for the two zone-specific models. Two radiologists (A and B) independently Likert scored test 3 T mp-MRI dataset, allowing ROC analysis for each radiologist for each prostate zone.ResultsDiagnostic models applied to the test dataset demonstrated a ROC-AUC = 0.74 (95% CI 0.67–0.81) in the PZ and 0.68 (95% CI 0.61–0.75) in the TZ. Radiologist A/B had a ROC-AUC = 0.78/0.74 in the PZ and 0.69/0.69 in the TZ. Radiologists A and B each scored 51 patients in the PZ and 41 and 45 patients respectively in the TZ as Likert 3. The PZ model demonstrated a ROC-AUC = 0.65/0.67 for the patients Likert scored as indeterminate by radiologist A/B respectively, whereas the TZ model demonstrated a ROC-AUC = 0.74/0.69.ConclusionZone-specific mp-MRI diagnostic models demonstrate generalisability between 1.5 and 3 T mp-MRI protocols and show similar classification performance to experienced radiologists for prostate cancer detection. Results also indicate the ability of diagnostic models to classify cases with an indeterminate radiologist score.Key Points• MRI diagnostic models had similar performance to experienced radiologists for classification of prostate cancer.• MRI diagnostic models may help radiologists classify tumour in patients with indeterminate Likert 3 scores.

Emergency Medicine Resident versus Radiologist in Detecting the Ultrasonographic Signs of Acute Cholecystitis; a Diagnostic Accuracy Study.

Dependence of ultrasonography on the operator's skill plays a major role in the differences between various studies in reporting its diagnostic accuracy. Therefore, the present study was done with the aim of comparing the ultrasonography findings performed by emergency medicine resident and radiologist in evaluation of acute cholecystitis. The present diagnostic accuracy study has been carried out on patients presenting to the emergency department with complaint of pain in the right upper quadrant of abdomen suspected with acute cholecystitis. All the patients underwent gallbladder ultrasonography by a trained emergency medicine resident and a radiologist and their findings were compared with surgical and pathology findings regarding gallstone and increased gallbladder wall thickness. 51 patients with the mean age of 42.3±15.8 (17-81) years were analyzed (82.4% female). The overall agreement between emergency medicine resident and radiologist in ultrasonographic diagnosis of cholecystitis was 0.421 (95% CI: 0.118-0.724). Based on the pathology and surgical findings, acute cholecystitis was confirmed for all 51 (100%) patients. Meanwhile, based on the ultrasonographic report of radiologist and emergency medicine resident only 45 (88.2%) and 34 (66.7%) patients, respectively, were diagnosed with cholecystitis. Screening performance characteristics of ultrasonography by radiologist for detection of gallbladder stone (p = 0.010) and gallbladder wall thickness (p < 0.0001) were significantly better than emergency medicine resident. The screening performance characteristics of ultrasonography by radiologist in detection of gallstones and increased wall thickness of gallbladder were significantly better.

Imaging Findings in Elder Abuse: A Role for Radiologists in Detection

PurposeEmergency department assessment represents a critical but often missed opportunity to identify elder abuse, which is common and has serious consequences. Among emergency care providers, diagnostic radiologists are optimally positioned to raise suspicion for mistreatment when reviewing imaging of geriatric injury victims. However, little literature exists describing relevant injury patterns, and most radiologists currently receive neither formal nor informal training in elder abuse identification. MethodsWe present 2 cases to begin characterisation of the radiographic findings in elder abuse. ResultsFindings from these cases demonstrate similarities to suspicious findings in child abuse including high-energy fractures that are inconsistent with reported mechanisms and the coexistence of acute and chronic injuries. Specific injuries uncommon to accidental injury are also noted, including a distal ulnar diaphyseal fracture. ConclusionsWe hope to raise awareness of elder abuse among diagnostic radiologists to encourage future large-scale research, increased focus on chronic osseous findings, and the addition of elder abuse to differential diagnoses.

Diagnostic Performance of Fused Diffusion-Weighted Imaging Using Unenhanced or Postcontrast T1-Weighted MR Imaging in Patients With Breast Cancer.

To evaluate the diagnostic performance of fused diffusion-weighted imaging (DWI) using either unenhanced (UFMR) or early postcontrast T1-weighted imaging (PCFMR) to detect and characterize breast lesions in patients with breast cancer.This retrospective observational study was approved by institutional review board in our hospital and informed consents were waived. We retrospectively selected 87 consecutive patients who underwent preoperative breast magnetic resonance imaging, including DWI and definitive surgery. Both UFMR and PCFMR were reviewed by 5 radiologists for detection, lesion size, Breast Imaging Reporting and Data System final assessment, the probability of malignancy, lesion conspicuity, and apparent diffusion coefficients.A total of 129 lesions were identified by at least 2 readers on UFMR or PCFMR. Of 645 potentially detected lesions, there were 528 (82%) with UFMR and 554 (86%) with PCFMR. Malignant lesions or index cancers showed significantly higher detection rates than benign or additional lesions on both UFMR and PCFMR (P < 0.05). Area under the characteristic curves (AUCs) for predicting malignancy ranged 0.927 to 0.986 for UFMR, and 0.936 to 0.993 for PCFMR, which was not significantly different. Lesion conspicuity was significantly higher on PCFMR than UFMR (8.59 ± 1.67 vs 9.19 ± 1.36, respectively; P < 0.05) across 5 readers. Mean intraclass correlation coefficients for lesion size on UFMR and PCFMR were 0.89 and 0.92, respectively.Detection rates of index malignant lesions were similar for UFMR and PCFMR. Interobserver agreement for final assessments was reliable across 5 readers. Diagnostic accuracy for predicting malignancy with UFMR versus PCFMR was similar, although lesion conspicuity was significantly greater with the latter.

English

Breast cancer is one of the most important reasons of death within women 15 to 54 years old. A woman dies of breast cancer every 13 minutes and 12.6 percent of women are infected during their life. Although mammography is one of the most effective ways of detecting disease; but still it has deficiencies and limitations. Cancerous lesions may not be diagnosed or non-cancerous lesions are detected as cancer in the interpretation of mammography images. Recently, quality of mammography images is increased and image features are extracted using image processing science in order to help radiologists in detection and diagnosis of cancer masses. This will increase detection speed and accuracy rate. In this study, a new method is proposed for detection of suspicious areas of breast cancer tumoursbased on wavelet and hidden Markov model. According to combination of these two methods, the efficiency has been increased compared to method of previous works. In this research, cancer masses are detected as well as percentage of cancer masses becomes clear; this makes to estimate mass growth rate. In this paper, Markov model with tree structure is used in order to extract statistical properties of wavelet transform components. Markov model has special ability in extracting information related to edges and protruding parts of image context due to its features which can accurately detect cancer areas. In this research we try to estimate the appropriate label (clustering) of pixels from a checked image in order to segment cancer areas. Certain joint distribution is assumed for pixels of a region or class; then, the maximum similarity of different areas of an image under review is checked using ML method. Combination of MIAS database and Paden including 150 images is used in order to test the proposed method. The results indicate that proposed method is more accurate compared to methods that use only wavelet transforms method. Detection rate of proposed method is 96% that is improved 24.5% compared to wavelet transform with detection rate of 71.5 percent.

Ultrasound Imaging of Nodal Metastases in the Central Compartment in Thyroid Cancer: is it Possible to Increase Sensitivity

Background and Purpose: Ultrasound imaging criteria for cervical lymph nodes were set mainly to increase sensitivity. However, unlike the lateral neck compartment, the challenge in the central compartment is the low sensitivity of Ultrasonography. We performed this study to evaluate whether the sensitivity of Ultrasonography could be increased by considering any detected lymph node in the central compartment as malignant. Methods: Forty one patients with differentiated thyroid cancer underwent dedicated pre-operative Ultrasonography by one radiologist for detection of metastatic lymph nodes in the central compartment. Any lymph node detected was considered positive regardless of Ultrasound morphologic criteria. All patients underwent thyroidectomy with central node dissection (CND), unilateral in 31 and bilateral in 10 cases, based on tumor size and laterality. Ultrasound imaging results were compared to histopathologic results. Results: The number of true positive cases by Ultrasonography was 13, false negative cases 8, true negative 15 and false positive 5 cases. The sensitivity of US was 61.9%, the specificity 75%, the positive predictive value 72.2% and the negative predictive value 65.2%. Conclusions: The sensitivity of Ultrasonography for detection of metastatic lymph nodes in the central compartment in patients with DTC may be increased, compared to previously published data, by considering every detected node as positive rather than using the routine criteria for lymph node metastases in other neck levels.

Predicting Arterial Injuries after Penetrating Brain Trauma Based on Scoring Signs from Emergency CT Studies

The objective of this study was to determine the accuracy of individual radiologists in detection of vascular injury in patients after penetrating brain injury (PBI) based on head CT findings at admission. We retrospectively evaluated 54 PBI patients who underwent admission head CT and digital subtraction angiography (DSA), used here as a reference standard. Two readers reviewed the CT images to determine the presence or absence of the 29 CT variables of injury profile and quantified selected variables. Four experienced trauma radiologists and one neuroradiologist assigned their own specific scores for each CT variable, a high score indicative of a high probability of artery injury. A sixth set consisted of the average score obtained from the five sets, generated by five experts. Receiver operating characteristic (ROC) curves were constructed for each set to assess the diagnostic performance of an individual radiologist in predicting an underlying vascular injury. The area under ROC curve (AUC) was higher for CT scores obtained from the sixth set (average of five sets of scores) of variable rank score 0.75 (95% CI 0.62-0.88) and for the rest of the data sets, the value ranged from 0.70 (95% CI 0.56-0.84) to 0.74 (95% CI 0.6-0.88). In conclusion, radiologists may be able to recommend DSA with a fair accuracy rate in selected patients, deemed 'high-risk' for developing intracranial vascular injuries after PBI based on admission CT studies. A better approach needs to be developed to reduce the false positive rate to avoid unnecessary emergency DSA.

Breast Cancer Diagnosis System Based on Wavelet Analysis and Neural Networks

The high incidence of breast cancer has increased significantly in the recent years. The most familiar breast tumors types are mass and microcalcifications (Mcs). Mammogram is considered the most reliable method in early detection of breast cancer. Computer-aided diagnosis system can be very helpful for radiologist in detection and diagnosing abnormalities earlier and faster than traditional screening programs. Several techniques can be used to accomplish this task. In this work, the authors present a preprocessing method, based on homomorphic filtering and wavelet, to extract the abnormal Mcs in mammographic images. The authors use four different methods of feature extraction for classification of normal and abnormal patterns in mammogram. Four different feature extraction methods are used here are Wavelet, Gist, Gabor and Tamura. A classification system based on neural network and nearest neighbor classification is used.

Correlated diffusion imaging

BackgroundProstate cancer is one of the leading causes of cancer death in the male population. Fortunately, the prognosis is excellent if detected at an early stage. Hence, the detection and localization of prostate cancer is crucial for diagnosis, as well as treatment via targeted focal therapy. New imaging techniques can potentially be invaluable tools for improving prostate cancer detection and localization.MethodsIn this study, we introduce a new form of diffusion magnetic resonance imaging called correlated diffusion imaging, where the tissue being imaged is characterized by the joint correlation of diffusion signal attenuation across multiple gradient pulse strengths and timings. By taking into account signal attenuation at different water diffusion motion sensitivities, correlated diffusion imaging can provide improved delineation between cancerous tissue and healthy tissue when compared to existing diffusion imaging modalities.ResultsQuantitative evaluation using receiver operating characteristic (ROC) curve analysis, tissue class separability analysis, and visual assessment by an expert radiologist were performed to study correlated diffusion imaging for the task of prostate cancer diagnosis. These results are compared with that obtained using T2-weighted imaging and standard diffusion imaging (via the apparent diffusion coefficient (ADC)). Experimental results suggest that correlated diffusion imaging provide improved delineation between healthy and cancerous tissue and may have potential as a diagnostic tool for cancer detection and localization in the prostate gland.ConclusionsA new form of diffusion magnetic resonance imaging called correlated diffusion imaging (CDI) was developed for the purpose of aiding radiologists in cancer detection and localization in the prostate gland. Preliminary results show CDI shows considerable promise as a diagnostic aid for radiologists in the detection and localization of prostate cancer.

"Mammographic Image Enhancement, Classification and Retrieval using Color, Statistical and Spectral Analysis"

One of the major causes of cancer death among middle aged women in developed countries is breast cancer. Mammography is one method used by radiologists for detection and interpretation of cancer in breast images. Over the past few years Content-based Image Retrieval (CBIR) approaches has received significant attention for medical images analysis. In this paper content based retrieval techniques were tested for tissue classification and analysis of breast images. The proposed method employs image enhancement, analysis and classification of mammograms using histogram, statistical, wavelet coefficients and spectral features. The implementation of proposed method was carried out using MATLAB software and hence can work on simple personal computer. Analysis was carried out on 56 images collected from open source mini-MIAS database. Euclidean distance was used to compare the features of query image with stored images in database. Results show that the suggested features can be used for both classification and retrieval of mammographic images. The retrieval efficiency was obtained to be 85.7%. General Terms Pattern Recognition, Medical Image Processing.