Lung-DETR: Deformable Detection Transformer for Sparse Lung Nodule Anomaly Detection

Hooman Ramezani

Keywords: Lung Nodule Detection, Anomaly Detection, Deformable Detection Transformer, Medical Imaging, Computed Tomography (CT), Sparse Data, Class Imbalance, Deep Learning in Healthcare, Focal Loss, Maximum Intensity Projection, Object Detection, Transformer Architecture, Lung Cancer Detection, Data Preprocessing, Clinical Application, Neural Networks

TL;DR: A novel method for lung tumor detection in CT scans combining Deformable-DETR, Focal Loss, and Maximum Intensity Projection into a unified framework, achieving a 94.2% F1 score in detecting sparse nodules in real-world clinical dataset.

Abstract: Accurate lung nodule detection for computed tomography (CT) scan imagery is challenging in real-world settings due to the sparse occurrence of nodules and similarity to other anatomical structures. In a typical positive case, nodules may appear in as few as 3\% of CT slices, complicating detection. To address this, we reframe the problem as an anomaly detection task, targeting rare nodule occurrences in a predominantly normal dataset. We introduce a novel solution leveraging custom data preprocessing and Deformable Detection Transformer (Deformable-DETR). A 7.5mm Maximum Intensity Projection (MIP) is utilized to combine adjacent lung slices into single images, reducing the slice count and decreasing nodule sparsity. This enhances spatial context, allowing for better differentiation between nodules and other structures such as complex vascular structures and bronchioles. Deformable-DETR is employed to detect nodules, with a custom focal loss function to better handle the imbalanced dataset. Our model achieves state-of-the-art performance on the LUNA16 dataset with an F1 score of 94.2\% (95.2\% recall, 93.3\% precision) on a dataset sparsely populated with lung nodules that is reflective of real-world clinical data.

Submission Number: 51