Deep learning-based classification of blue light cystoscopy imaging during transurethral resection of bladder tumors

Abstract

Bladder cancer is one of the top 10 frequently occurring cancers and leads to most cancer deaths worldwide. Recently, blue light (BL) cystoscopy-based photodynamic diagnosis was introduced as a unique technology to enhance the detection of bladder cancer, particularly for the detection of flat and small lesions. Here, we aim to demonstrate a BL image-based artificial intelligence (AI) diagnostic platform using 216 BL images, that were acquired in four different urological departments and pathologically identified with respect to cancer malignancy, invasiveness, and grading. Thereafter, four pre-trained convolution neural networks were utilized to predict image malignancy, invasiveness, and grading. The results indicated that the classification sensitivity and specificity of malignant lesions are 95.77% and 87.84%, while the mean sensitivity and mean specificity of tumor invasiveness are 88% and 96.56%, respectively. This small multicenter clinical study clearly shows the potential of AI based classification of BL images allowing for better treatment decisions and potentially higher detection rates.

Advances in Diagnosis and Therapy for Bladder Cancer

“…Shkolyar et al developed a convolutional neural network-based image analysis platform (CystoNet) for the automatic detection of BCa with a sensitivity of 90.9% (95% CI: 90.3-91.6%) and a specificity of 98.6% (95% CI: 98.5-98.8%) [42]. Ali et al introduced an AI diagnostic platform, which depended on four pre-trained convolutional neural networks (CNN) to predict the malignancy, invasiveness, and grading of the images with a sensitivity of 95.77% and a specificity of 87.84%, respectively [43]. The classification pipeline for AI detection of malignant tumor task and the detailed process of it is illustrated in Figure 2.…”

Active surveillance for non-muscle-invasive bladder cancer: fallacy or opportunity?

“…Aside from critical patient selection, technological advances using artificial intelligence (AI)-and deep learning-based tools might help overcome the beforementioned inaccuracies in staging and improve surveillance of NMIBC [72][73][74][75]. Although these technologies are not yet available in practice, they have shown to outperform clinicians and might become a game-changer in the future AS strategy [75].…”

A Comparative Study of Deep Neural Networks for Real-Time Semantic Segmentation during the Transurethral Resection of Bladder Tumors

“…Bladder cancer is one of the ten most common types of cancer and is also reported to be a leading cause of death in Western countries [ 1 ]. The diagnosis is most often based on white-light cystoscopy imaging, which is excellent for the detection of papillary bladder tumors, but small or flat lesions may not be visible clearly enough [ 2 ]. Furthermore, the interpretation of cystoscopy findings can vary from examiner to examiner, since no two physicians have the same skills and experience [ 3 ].…”

“…Deep learning (DL) methods are developing rapidly, and as a result, neural network-based bladder cancer diagnosis is becoming increasingly popular [ 4 , 5 , 6 ]. DL solutions can not only achieve diagnostic accuracy similar to that of experienced specialists [ 2 , 7 ], but their results are also objective and reproducible [ 3 ], as they are based on mathematical operations. Using neural networks to help diagnose bladder cancer could result in fewer tumors going undiagnosed, which could save lives.…”

“…Another example is the work of Ali et al [ 2 ], who proposed an artificial intelligence diagnostic platform that operated on high-quality blue-light cystoscopy images and predicted cancer malignancy, invasiveness, and grade. In their platform, they implemented four convolutional neural networks (VGG-16 [ 11 ], ResNet-50 [ 12 ], MobileNetV2 [ 13 ] and InceptionV3 [ 14 ]), which were pre-trained and fine-tuned via transfer learning.…”