Automatic renal carcinoma biopsy guidance using forward-viewing endoscopic optical coherence tomography and deep...

Area of Science:

• Medical Imaging
• Oncology
• Biotechnology

Background:

• Percutaneous renal biopsy is crucial for kidney cancer diagnosis but faces challenges in sampling accuracy.
• Differentiating between malignant and benign renal tumors requires precise tissue identification.

Purpose of the Study:

• To develop and evaluate a forward-viewing optical coherence tomography (OCT) probe for real-time differentiation of tumor and normal renal tissues.
• To enhance the precision of biopsy guidance in kidney cancer diagnosis.
• To assess the capability of convolutional neural networks (CNNs) in classifying renal tissues.

Main Methods:

• A forward-viewing OCT probe was utilized for in vivo imaging of human kidney samples.
• Distinct imaging features of carcinoma, oncocytoma, and normal renal tissues were analyzed.
• Convolutional neural networks (CNNs) were developed and trained for automated tissue recognition.
• Performance was compared against the conventional attenuation coefficient method.

Main Results:

• The OCT probe successfully distinguished carcinoma from normal renal tissues based on unique imaging characteristics.
• Oncocytoma was effectively differentiated from carcinoma.
• CNN models achieved a high tissue recognition accuracy of 99.1% on a hold-out dataset.
• The CNN-aided platform demonstrated superior accuracy in carcinoma prediction compared to the attenuation coefficient method.

Conclusions:

• The developed CNN-aided endoscopic imaging platform significantly enhances diagnostic accuracy during percutaneous renal biopsy.
• This technology offers a promising tool for precise guidance in kidney cancer diagnosis and characterization.
• Real-time tissue differentiation capabilities can improve biopsy sampling and patient outcomes.