Deep learning model for predicting the presence of stromal invasion of breast cancer on digital breast tomosynthesis.

To develop a deep learning (DL)-based algorithm to predict the presence of stromal invasion in breast cancer using digital breast tomosynthesis (DBT). Our institutional review board approved this retrospective study and waived the requirement for informed consent from the patients. Initially, 499 patients (mean age 50.5 years, age range, 29-90 years) who were referred to our hospital under the suspicion of breast cancer and who underwent DBT between March 1 and August 31, 2019, were enrolled in this study. Among the 499 patients, 140 who underwent surgery after being diagnosed with breast cancer were selected for the analysis. Based on the pathological reports, the 140 patients were classified into two groups: those with non-invasive cancer (n = 20) and those with invasive cancer (n = 120). VGG16, Resnet50, DenseNet121, and Xception architectures were used as DL models to differentiate non-invasive from invasive cancer. The diagnostic performance of the DL models was assessed based on the area under the receiver operating characteristic curve (AUC). The AUC for the four models were 0.56 [95% confidence intervals (95% CI) 0.49-0.62], 0.67 (95% CI 0.62-0.74), 0.71 (95% CI 0.65-0.75), and 0.75 (95% CI 0.69-0.81), respectively. Our proposed DL model trained on DBT images is useful for predicting the presence of stromal invasion in breast cancer.

Automatic breast carcinoma detection in histopathological micrographs based on Single Shot Multibox Detector.

Background: A diagnosis with histological classification by pathologists is very important for appropriate treatments to improve the prognosis of patients with breast cancer. However, the number of pathologists is limited, and assisting the pathological diagnosis by artificial intelligence becomes very important. Here, we presented an automatic breast lesions detection model using microscopic histopathological images based on a Single Shot Multibox Detector (SSD) for the first time and evaluated its significance in assisting the diagnosis. Methods: We built the data set and trained the SSD model with 1361 microscopic images and evaluated using 315 images. Pathologists and medical students diagnosed the images with or without the assistance of the model to investigate the significance of our model in assisting the diagnosis. Results: The model achieved 88.3% and 90.5% diagnostic accuracies in 3-class (benign, non-invasive carcinoma, or invasive carcinoma) or 2-class (benign or malignant) classification tasks, respectively, and the mean intersection over union was 0.59. Medical students achieved a remarkably higher diagnostic accuracy score (average 84.7%) with the assistance of the model compared to those without assistance (average 67.4%). Some people diagnosed images in a short time using the assistance of the model (shorten by average 6.4 min) while others required a longer time (extended by 7.2 min). Conclusion: We presented the automatic breast lesions detection method at high speed using histopathological micrographs. The present system may conveniently support the histological diagnosis by pathologists in laboratories.