Automated detection and segmentation of thoracic lymph nodes from CT using 3D foveal fully convolutional neural networks

Abstract

Background In oncology, the correct determination of nodal metastatic disease is essential for patient management, as patient treatment and prognosis are closely linked to the stage of the disease. The aim of the study was to develop a tool for automatic 3D detection and segmentation of lymph nodes (LNs) in computed tomography (CT) scans of the thorax using a fully convolutional neural network based on 3D foveal patches. Methods The training dataset was collected from the Computed Tomography Lymph Nodes Collection of the Cancer Imaging Archive, containing 89 contrast-enhanced CT scans of the thorax. A total number of 4275 LNs was segmented semi-automatically by a radiologist, assessing the entire 3D volume of the LNs. Using this data, a fully convolutional neuronal network based on 3D foveal patches was trained with fourfold cross-validation. Testing was performed on an unseen dataset containing 15 contrast-enhanced CT scans of patients who were referred upon suspicion or for staging of bronchial carcinoma. Results The algorithm achieved a good overall performance with a total detection rate of 76.9% for enlarged LNs during fourfold cross-validation in the training dataset with 10.3 false-positives per volume and of 69.9% in the unseen testing dataset. In the training dataset a better detection rate was observed for enlarged LNs compared to smaller LNs, the detection rate for LNs with a short-axis diameter (SAD) ≥ 20 mm and SAD 5–10 mm being 91.6% and 62.2% (p < 0.001), respectively. Best detection rates were obtained for LNs located in Level 4R (83.6%) and Level 7 (80.4%). Conclusions The proposed 3D deep learning approach achieves an overall good performance in the automatic detection and segmentation of thoracic LNs and shows reasonable generalizability, yielding the potential to facilitate detection during routine clinical work and to enable radiomics research without observer-bias.

Deep Learning-Based Segmentation and Volume Calculation of Pediatric Lymphoma on Contrast-Enhanced Computed Tomographies

“…Early attempts employed numerical methods analyzing the shape, gray value, and borders of the nodes that generally required placement of a marker inside a lymph node that was to be segmented [ 34 ], although some models were also able to detect lymph nodes [ 35 ] without placing the marker. More recently, Iuga et al [ 36 ] proposed a CNN for the detection and segmentation of thoracic lymph nodes in patients with possible lymph nodes metastasis. However, their approach differed from ours, as multiple small, non-pathological lymph nodes were generally assessed, and their primary benchmark was the number of detected lymph nodes and not their volume.…”

Combination of Whole-Body Baseline CT Radiomics and Clinical Parameters to Predict Response and Survival in a Stage-IV Melanoma Cohort Undergoing Immunotherapy

“…Although the clinical application of a potential biomarker would be challenging with a whole-body segmentation approach, the study’s aim was to use as much information about the metastatic load as possible to create a benchmark for predictive performance. Furthermore, due to modern (semi)automated segmentation tools, whole-body segmentation will become faster and easier to apply [ 38 , 39 , 40 ].…”

Artificial Intelligence and Precision Medicine: A New Frontier for the Treatment of Brain Tumors

“…In medicine, AI is used for automated diagnostic procedures and treatments for patients [ 119 ]. An AI-based approach to cancer imaging can help improve tumor detection and characterization, as well as monitor the tumor’s response to treatment and check for early signs of cancer in other parts of the body [ 120 ].…”