Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Detection of adrenal anomalous findings in spinal CT images using multi model graph aggregation

Carmel Shabalin, Israel Shenkman, Ilan Shelef, Gal Ben-Arie, Alex Geftler, Yuval Shahar

TL;DR

This work addresses the problem of incidental adrenal anomaly detection on spine-focused CT scans by introducing the Multi Model Graph Aggregation (MMGA) pipeline, which combines slice filtering (CNN), adrenal detection (YOLOv3), and graph-based aggregation (DGCNN) to produce patient-level classification and localization. The study demonstrates that MMGA can achieve a PPV of up to 0.67, NPV of 0.944, and localization IOUs around 0.41 (left) and 0.52 (right), with a ROC-AUC up to 0.869, highlighting its potential to reduce radiologist workload while improving detection. Key contributions include leveraging spine-oriented CT for abdominal anomaly screening, a three-model architecture that can generalize to other organs and imaging types, and a graph-based aggregation strategy that effectively fuses multi-slice, multi-model detections. The approach offers practical impact for automated screening in clinical workflows and paves the way for multi-modal extensions and broader organ coverage.

Abstract

Low back pain is the symptom that is the second most frequently reported to primary care physicians, effecting 50 to 80 percent of the population in a lifetime, resulting in multiple referrals of patients suffering from back problems, to CT and MRI scans, which are then examined by radiologists. The radiologists examining these spinal scans naturally focus on spinal pathologies and might miss other types of abnormalities, and in particular, abdominal ones, such as malignancies. Nevertheless, the patients whose spine was scanned might as well have malignant and other abdominal pathologies. Thus, clinicians have suggested the need for computerized assistance and decision support in screening spinal scans for additional abnormalities. In the current study, We have addressed the important case of detecting suspicious lesions in the adrenal glands as an example for the overall methodology we have developed. A patient CT scan is integrated from multiple slices with an axial orientation. Our method determines whether a patient has an abnormal adrenal gland, and localises the abnormality if it exists. Our method is composed of three deep learning models; each model has a different task for achieving the final goal. We call our compound method the Multi Model Graph Aggregation MMGA method. The novelty in this study is twofold. First, the use, for an important screening task, of CT scans that are originally focused and tuned for imaging the spine, which were acquired from patients with potential spinal disorders, for detection of a totally different set of abnormalities such as abdominal Adrenal glands pathologies. Second, we have built a complex pipeline architecture composed from three deep learning models that can be utilized for other organs (such as the pancreas or the kidney), or for similar applications, but using other types of imaging, such as MRI.

Detection of adrenal anomalous findings in spinal CT images using multi model graph aggregation

TL;DR

This work addresses the problem of incidental adrenal anomaly detection on spine-focused CT scans by introducing the Multi Model Graph Aggregation (MMGA) pipeline, which combines slice filtering (CNN), adrenal detection (YOLOv3), and graph-based aggregation (DGCNN) to produce patient-level classification and localization. The study demonstrates that MMGA can achieve a PPV of up to 0.67, NPV of 0.944, and localization IOUs around 0.41 (left) and 0.52 (right), with a ROC-AUC up to 0.869, highlighting its potential to reduce radiologist workload while improving detection. Key contributions include leveraging spine-oriented CT for abdominal anomaly screening, a three-model architecture that can generalize to other organs and imaging types, and a graph-based aggregation strategy that effectively fuses multi-slice, multi-model detections. The approach offers practical impact for automated screening in clinical workflows and paves the way for multi-modal extensions and broader organ coverage.

Abstract

Low back pain is the symptom that is the second most frequently reported to primary care physicians, effecting 50 to 80 percent of the population in a lifetime, resulting in multiple referrals of patients suffering from back problems, to CT and MRI scans, which are then examined by radiologists. The radiologists examining these spinal scans naturally focus on spinal pathologies and might miss other types of abnormalities, and in particular, abdominal ones, such as malignancies. Nevertheless, the patients whose spine was scanned might as well have malignant and other abdominal pathologies. Thus, clinicians have suggested the need for computerized assistance and decision support in screening spinal scans for additional abnormalities. In the current study, We have addressed the important case of detecting suspicious lesions in the adrenal glands as an example for the overall methodology we have developed. A patient CT scan is integrated from multiple slices with an axial orientation. Our method determines whether a patient has an abnormal adrenal gland, and localises the abnormality if it exists. Our method is composed of three deep learning models; each model has a different task for achieving the final goal. We call our compound method the Multi Model Graph Aggregation MMGA method. The novelty in this study is twofold. First, the use, for an important screening task, of CT scans that are originally focused and tuned for imaging the spine, which were acquired from patients with potential spinal disorders, for detection of a totally different set of abnormalities such as abdominal Adrenal glands pathologies. Second, we have built a complex pipeline architecture composed from three deep learning models that can be utilized for other organs (such as the pancreas or the kidney), or for similar applications, but using other types of imaging, such as MRI.

Paper Structure

This paper contains 38 sections, 3 equations, 23 figures, 6 tables.

Table of Contents

  1. INTRODUCTION
  2. Motivation
  3. Some reaserch background
  4. Methodology and Novelty
  5. DOMAIN BACKGROUND
  6. Abdominal pathologies and anatomy
  7. Basic abdominal anatomy relevant to adrenal glands
  8. Common abdominal diseases and prevalence in population
  9. Overview on Convolutional Neural Networks
  10. Overview on Object detection
  11. Overview on Graph Learning
  12. Computer-Aided Diagnosis
  13. Automatic ventral anomaly detection
  14. METHODS
  15. Data preprocessing
  16. ...and 23 more sections

Figures (23)

  • Figure 1: Adrenal glands zoom in
  • Figure 2: Left : Artificial Neuron (processing unit); Right: ANN architecture
  • Figure 3: Convolutional Neural Network basic architecture
  • Figure 4: High level diagram of YOLO framework for generic object detection redmon2016you
  • Figure 5: The compound hierarchical segmentation architecture. The left-most U-Net segments the lungs’ cavity and the right one partitions the lung area into GGO, CON, and healthy tissue ben2022deep.
  • ...and 18 more figures