Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Weakly Supervised Detection of Pheochromocytomas and Paragangliomas in CT

David C. Oluigboa, Bikash Santra, Tejas Sudharshan Mathai, Pritam Mukherjee, Jianfei Liu, Abhishek Jha, Mayank Patel, Karel Pacak, Ronald M. Summers

TL;DR

This work tackles the challenge of detecting pheochromocytomas and paragangliomas (PPGLs) in CT scans using weak annotations. It introduces a proxy segmentation framework where 2D radiologist-marked boxes are expanded to weak 3D masks and combined with body-region segmentation, trained with a 3D full-resolution nnUNet. The method achieves a precision of $62.4 ext{ extpercent}$ and recall of $64.1 ext{ extpercent}$ without size filtering, improving to $70.0 ext{ extpercent}$ precision when excluding small predictions, with a patient-level recall of $100 ext{ extpercent}$ and median precision up to $84.9 ext{ extpercent}$. This pilot demonstrates the feasibility of weakly supervised PPGL detection in CT and paves the way for more efficient, automated monitoring of PPGLs in clinical workflows, albeit with limitations such as ground-truth bias and lack of genetic stratification.

Abstract

Pheochromocytomas and Paragangliomas (PPGLs) are rare adrenal and extra-adrenal tumors which have the potential to metastasize. For the management of patients with PPGLs, CT is the preferred modality of choice for precise localization and estimation of their progression. However, due to the myriad variations in size, morphology, and appearance of the tumors in different anatomical regions, radiologists are posed with the challenge of accurate detection of PPGLs. Since clinicians also need to routinely measure their size and track their changes over time across patient visits, manual demarcation of PPGLs is quite a time-consuming and cumbersome process. To ameliorate the manual effort spent for this task, we propose an automated method to detect PPGLs in CT studies via a proxy segmentation task. As only weak annotations for PPGLs in the form of prospectively marked 2D bounding boxes on an axial slice were available, we extended these 2D boxes into weak 3D annotations and trained a 3D full-resolution nnUNet model to directly segment PPGLs. We evaluated our approach on a dataset consisting of chest-abdomen-pelvis CTs of 255 patients with confirmed PPGLs. We obtained a precision of 70% and sensitivity of 64.1% with our proposed approach when tested on 53 CT studies. Our findings highlight the promising nature of detecting PPGLs via segmentation, and furthers the state-of-the-art in this exciting yet challenging area of rare cancer management.

Weakly Supervised Detection of Pheochromocytomas and Paragangliomas in CT

TL;DR

This work tackles the challenge of detecting pheochromocytomas and paragangliomas (PPGLs) in CT scans using weak annotations. It introduces a proxy segmentation framework where 2D radiologist-marked boxes are expanded to weak 3D masks and combined with body-region segmentation, trained with a 3D full-resolution nnUNet. The method achieves a precision of and recall of without size filtering, improving to precision when excluding small predictions, with a patient-level recall of and median precision up to . This pilot demonstrates the feasibility of weakly supervised PPGL detection in CT and paves the way for more efficient, automated monitoring of PPGLs in clinical workflows, albeit with limitations such as ground-truth bias and lack of genetic stratification.

Abstract

Pheochromocytomas and Paragangliomas (PPGLs) are rare adrenal and extra-adrenal tumors which have the potential to metastasize. For the management of patients with PPGLs, CT is the preferred modality of choice for precise localization and estimation of their progression. However, due to the myriad variations in size, morphology, and appearance of the tumors in different anatomical regions, radiologists are posed with the challenge of accurate detection of PPGLs. Since clinicians also need to routinely measure their size and track their changes over time across patient visits, manual demarcation of PPGLs is quite a time-consuming and cumbersome process. To ameliorate the manual effort spent for this task, we propose an automated method to detect PPGLs in CT studies via a proxy segmentation task. As only weak annotations for PPGLs in the form of prospectively marked 2D bounding boxes on an axial slice were available, we extended these 2D boxes into weak 3D annotations and trained a 3D full-resolution nnUNet model to directly segment PPGLs. We evaluated our approach on a dataset consisting of chest-abdomen-pelvis CTs of 255 patients with confirmed PPGLs. We obtained a precision of 70% and sensitivity of 64.1% with our proposed approach when tested on 53 CT studies. Our findings highlight the promising nature of detecting PPGLs via segmentation, and furthers the state-of-the-art in this exciting yet challenging area of rare cancer management.
Paper Structure (7 sections, 2 figures, 1 table)

This paper contains 7 sections, 2 figures, 1 table.

Table of Contents

  1. INTRODUCTION
  2. Methods
  3. Dataset
  4. Ground-Truth Generation
  5. Detection of PPGL using nnUNet
  6. Experiments and Results
  7. Discussion and Conclusion

Figures (2)

  • Figure 1: Framework for the detection of pheochromocytomas and paragangliomas (PPGLs) via a proxy segmentation task using a 3D nnUNet. PPGLs in CT volumes were annotated with 2D boxes (red box), and these were converted into weak 3D segmentations (yellow box). The body region mask (green) from TotalSegmentator was also merged with the weak 3D annotations. The 3D nnUNet model was trained to segment PPGLs annotated in the CT volumes. At test time, the model received a 3D CT volume and detected PPGLs (via segmentation).
  • Figure 2: Rows 1, 2, and 3 show cropped CT slices, ground-truth PPGLs (yellow), and detected PPGLs (blue) overlaid, respectively. Columns 1, 2, and 3 show true positives; a small part of an oblong tumor in column 3 was detected by nnUNet. In Column 4, a false positive incorrectly predicted by nnUNet is shown.