Reproducibility and Explainability of Deep Learning in Mammography: A Systematic Review of Literature.

Background Although abundant literature is currently available on the use of deep learning for breast cancer detection in mammography, the quality of such literature is widely variable. Purpose To evaluate published literature on breast cancer detection in mammography for reproducibility and to ascertain best practices for model design. Methods The PubMed and Scopus databases were searched to identify records that described the use of deep learning to detect lesions or classify images into cancer or noncancer. A modification of Quality Assessment of Diagnostic Accuracy Studies (mQUADAS-2) tool was developed for this review and was applied to the included studies. Results of reported studies (area under curve [AUC] of receiver operator curve [ROC] curve, sensitivity, specificity) were recorded. Results A total of 12,123 records were screened, of which 107 fit the inclusion criteria. Training and test datasets, key idea behind model architecture, and results were recorded for these studies. Based on mQUADAS-2 assessment, 103 studies had high risk of bias due to nonrepresentative patient selection. Four studies were of adequate quality, of which three trained their own model, and one used a commercial network. Ensemble models were used in two of these. Common strategies used for model training included patch classifiers, image classification networks (ResNet in 67%), and object detection networks (RetinaNet in 67%). The highest reported AUC was 0.927 ± 0.008 on a screening dataset, while it reached 0.945 (0.919-0.968) on an enriched subset. Higher values of AUC (0.955) and specificity (98.5%) were reached when combined radiologist and Artificial Intelligence readings were used than either of them alone. None of the studies provided explainability beyond localization accuracy. None of the studies have studied interaction between AI and radiologist in a real world setting. Conclusion While deep learning holds much promise in mammography interpretation, evaluation in a reproducible clinical setting and explainable networks are the need of the hour.

Imaging Spectrum in Chronic Pulmonary Aspergillosis.

Chronic pulmonary aspergillosis (CPA) is a life-threatening respiratory fungal infection that is almost exclusively seen in patients with preexisting structural lung disease with no or mild immunosuppression. The clinical presentation and imaging findings are varied and often pose a diagnostic challenge; and the disease is often present for a long time before being correctly diagnosed. High-resolution chest computed tomography is the imaging modality of choice because it helps identify various forms of CPA, which can range from a simple aspergilloma and chronic cavitary form, to the subacute invasive and end-stage fibrotic form. The knowledge of the imaging features of this disease cannot be overemphasized because it can assist the clinician in reaching at an early diagnosis and timely initiation of appropriate antifungal therapy, thereby improving patient management and treatment outcome. Moreover, imaging also plays a pivotal role during follow-up in patients of CPA to assess the treatment response. In the current review, we present an illustrative review of radiologic patterns seen in various forms of CPA.

Unusual Cause of Encephalopathy and Seizures in a Child.

Toxic encephalopathy is an important differential diagnosis in a child with encephalopathy and seizures. Subtle circumstantial evidence and apt neuroimaging features can significantly contribute to management, especially in a case of accidental exposure. 2, 4-D (ethyl ester) poisoning is a rare diagnosis, despite the common usage of this toxic compound as weedicide in northern India. The clinical similarity to the anticholinesterase poisoning, especially in the setting of agrochemical exposure is the main cause of under-diagnosis with usually fatal outcomes. We present an interesting case of accidental 2, 4-D (ethyl ester) poisoning in a child with typical neuroimaging features. A review of the literature regarding neuroimaging patterns of bilaterally symmetrical signal abnormalities involving basal ganglia in brain magnetic resonance imaging (MRI) from the point of view of clinical significance, is also discussed.

Evaluating awareness and practices pertaining to radioactive waste management among scrap dealers in Delhi, India.

OBJECTIVES: With nuclear technology rapidly taking the spotlight in the last 50 years, radiation accidents seem to be a harsh reality of the modern world. The Mayapuri Radiation accident of 2010 was the worst radiation accident India has yet dealt with. Two years thereafter, we designed a study to assess the awareness and practices regarding radioactive waste among scrap dealers aiming to assess deficiencies in radiation disaster preparedness. METHODOLOGY: A community based cross-sectional study. The study population consisted of 209 volunteers (from 108 scrap dealerships) including 108 shop-owners and 101 workers segregated as Group A consisting of 54 dealerships in Mayapuri and Group B of 54 dealerships from the rest of the city. Subjects were then interviewed using a semi-structured questionnaire. RESULTS: Awareness about radioactive waste varied significantly with level of education (p = 0.024), Kuppuswamy's socio-economic scale (p = 0.005), age of the scrap dealer (p = 0.049) and his work experience (p = 0.045). The larger dealerships in Mayapuri were more aware about radioactive waste (p = 0.0004), the accident in 2010 (p = 0.0002), the symbol for radiation hazard (p = 0.016), as well as the emergency guidelines and the agencies to contact in the event of a radiation accident. CONCLUSIONS: Our findings seem to signify that while governmental and non-governmental agencies were successful in implementing prompt disaster response and awareness programs, the community continues to be inadequately prepared. These go on to suggest that though concerted awareness and training programs do benefit the affected community, economic and social development is the key to disaster prevention and mitigation.