Classification of chest radiographs using general purpose cloud-based automated machine learning: pilot study

• Published in: Egyptian Journal of Radiology and Nuclear Medicine Vol. 52; no. 1; pp. 120 - 9
• Main Authors: Ghosh, Tamaghna, Tanwar, Simran, Chumber, Shishir, Vani, Kavita
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 05.05.2021; Springer; Springer Nature B.V; SpringerOpen
• Subjects: Accuracy; Algorithms; Automation; Cloud computing [L01.224.097]; Computer-assisted diagnosis [E01.158]; Datasets; Deep learning; Deep learning [G17.035.250.500.250]; Image retrieval; Imaging; Internet service providers; Machine learning; Machine learning [G17.035.250.500]; Medical imaging equipment; Medical research; Medicine; Medicine & Public Health; Neural networks; Neural networks [G17.485]; Nuclear Medicine; Radiography [E01.370.350.700]; Radiology; Semiconductor industry; Technology application; Tuberculosis; United States; United States > US; Radiography [E01.370.350.700]; Cloud computing [L01.224.097]; Computer-assisted diagnosis [E01.158]; Deep learning [G17.035.250.500.250]; Machine learning [G17.035.250.500]; Neural networks [G17.485]
• ISSN: 0378-603X; 2090-4762
• DOI: 10.1186/s43055-021-00499-w

Background Widespread implementation of machine learning models in diagnostic imaging is restricted by dearth of expertise and resources. General purpose automated machine learning offers a possible solution. This study aims to provide a proof of concept that a general purpose automated machine learning platform can be utilized to train a CNN to classify chest radiographs. In a retrospective study, more than 2000 postero-anterior chest radiographs were assessed for quality, contrast, position, and pathology. A selected dataset of 637 radiographs were used to train a CNN using reinforcement learning based automated machine learning platform. Accuracy metrics of each label was calculated and model performance was compared to previous studies. Results The auPRC (area under precision-recall curve) was 0.616. The model achieved precision of 70.8% and recall of 60.7% ( P > 0.05) for detection of “Normal” radiographs. Detection of “Pathology” by the model had a precision of 75.6% and recall of 75.6% ( P > 0.05). The F1 scores were 0.65 and 0.75 respectively. Conclusion Automated machine learning platforms may provide viable alternatives to developing custom CNN models for classification of chest radiographs. However, the accuracy achieved is lower than a comparable traditionally developed neural network model.