Deep learning for end-to-end kidney cancer diagnosis on multi-phase abdominal computed tomography

Abstract

In 2020, it is estimated that 73,750 kidney cancer cases were diagnosed, and 14,830 people died from cancer in the United States. Preoperative multi-phase abdominal computed tomography (CT) is often used for detecting lesions and classifying histologic subtypes of renal tumor to avoid unnecessary biopsy or surgery. However, there exists inter-observer variability due to subtle differences in the imaging features of tumor subtypes, which makes decisions on treatment challenging. While deep learning has been recently applied to the automated diagnosis of renal tumor, classification of a wide range of subtype classes has not been sufficiently studied yet. In this paper, we propose an end-to-end deep learning model for the differential diagnosis of five major histologic subtypes of renal tumors including both benign and malignant tumors on multi-phase CT. Our model is a unified framework to simultaneously identify lesions and classify subtypes for the diagnosis without manual intervention. We trained and tested the model using CT data from 308 patients who underwent nephrectomy for renal tumors. The model achieved an area under the curve (AUC) of 0.889, and outperformed radiologists for most subtypes. We further validated the model on an independent dataset of 184 patients from The Cancer Imaging Archive (TCIA). The AUC for this dataset was 0.855, and the model performed comparably to the radiologists. These results indicate that our model can achieve similar or better diagnostic performance than radiologists in differentiating a wide range of renal tumors on multi-phase CT.

Recognizing pathology of renal tumor from macroscopic cross-section image by deep learning

“…From the view of the comprehensive performance of all metrics and the statistical significance in AUC, the EfficientNet-B4 could be considered the best among three models for both of malignant and benign multi-class classification task, which revealed the advantage of EfficientNet-B4 and the certain consistency between two multi-class classification task. Besides, the previously reported studies about renal tumor diagnosis were mainly based on MRI or CT images and covered less subtypes compared with our study [27,29,33,35,39]. In the clinical practice, more renal tumor subtypes are desired to be diagnosed and the diagnosis process are expected to be as efficient as possible.…”

“…Considering that there is substantial overlap in the imaging findings of benign and malignant renal masses, Coy et al used both of deep learning and radiomics method to distinguish clear cell RCC from benign oncocytoma based on multiphasic CT images [28]. Kwang-Hyun et al proposed to identify five major histologic subtypes of renal tumors based on multi-phase CT using the end-to-end deep learning framework [29]. Xi et al developed a deep learning model to distinguish the benign tumors from RCC based on routine MR imaging [30].…”

Use of artificial intelligence to characterize renal tumors

“…AI showed better performance than the radiologists for diagnosis of the other types of RCC. In diagnosing benign tumors of oncocytoma and fat-poor AML, AI performed significantly better than the radiologists [ 6 ].…”

CT-based radiomics in predicting pathological response in non-small cell lung cancer patients receiving neoadjuvant immunotherapy

“…With the development of computer software, computational power has significantly improved, and in recent years, artificial intelligence (AI) technology based on deep learning (DL) algorithms has been vigorously developed and has gradually begun to be applied in medical research. Currently, it is mainly based on medical images using computer vision technology to solve clinical tasks such as lesion segmentation and disease classification ( 27 – 29 ). In the processing of medical images, the most widely used DL network is the convolutional neural network (CNN).…”