Classification of Multiple Diseases on Body CT Scans Using Weakly Supervised Deep Learning

Abstract

To design multidisease classifiers for body CT scans for three different organ systems using automatically extracted labels from radiology text reports. Materials and Methods:This retrospective study included a total of 12 092 patients (mean age, 57 years 6 18 [standard deviation]; 6172 women) for model development and testing. Rule-based algorithms were used to extract 19 225 disease labels from 13 667 body CT scans performed between 2012 and 2017. Using a three-dimensional DenseVNet, three organ systems were segmented: lungs and pleura, liver and gallbladder, and kidneys and ureters. For each organ system, a three-dimensional convolutional neural network classified each as no apparent disease or for the presence of four common diseases, for a total of 15 different labels across all three models. Testing was performed on a subset of 2158 CT volumes relative to 2875 manually derived reference labels from 2133 patients (mean age, 58 years 6 18; 1079 women). Performance was reported as area under the receiver operating characteristic curve (AUC), with 95% CIs calculated using the DeLong method.Results: Manual validation of the extracted labels confirmed 91%-99% accuracy across the 15 different labels. AUCs for lungs and pleura labels were as follows:

Annotation-free multi-organ anomaly detection in abdominal CT using free-text radiology reports: A multi-center retrospective study

“…We demonstrated that the models trained on a large amount of data labeled by a language model outperformed those trained on smaller datasets labeled by experts, as shown in Figure 6. Additionally, compared with a previous study that used reports as training labels, 24 our pipeline showed at least a 7% improvement in the AUC for organ-specific abnormality detection. This improvement may be due to the accuracy of the information extraction schema.…”

“…Only two studies to date have used reports as labels for CT images: one focused on the head region, with images collected from two institutions, 13 while the other targeted two abdominal regions (liver/gallbladder and kidney), using a total of 9,153 images from a single institution, which yielded an anomaly-detection performance of at least 7% lower than that of our pipeline. 24 Abnormality detection in abdominal CT images is challenging, because these scans provide high-resolution images of large areas including multiple organs. These images cannot be directly used in AI due to their size.…”

“…Only two studies to date have used reports as labels for CT images: one focused on the head region, with images collected from two institutions, 13 while the other targeted two abdominal regions (liver/gallbladder and kidney), using a total of 9,153 images from a single institution, which yielded an anomaly-detection performance of at least 7% lower than that of our pipeline. 24…”

Tailored 3D CT contrastive pretraining to improve pulmonary pathology classification

“…There has been substantial prior work on multipleabnormality prediction in different modalities; chest CT [1], body CT [2], chest x-ray for COVID-19 [3] and 3D MRI for brain tumor classification [4]. However, constructing multiclass (i.e., pathology) classifiers for 3D medical data isn't a straightforward undertaking and remains challenging because of two main obstacles: acquiring sufficiently large datasets and learning effective representations with weak pathological footprints.…”