Deep learning based tractography with TractSeg in patients with hemispherotomy: Evaluation and refinement
Deep learning-based tractography implicitly learns anatomical prior knowledge that is required to resolve ambiguities inherent in traditional streamline tractography. TractSeg is a particularly widely used example of such an approach. Even though it has exclusively been trained on healthy subjects,...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-01-01
|
| Series: | NeuroImage: Clinical |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2213158225000087 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1825199447658725376 |
|---|---|
| author | Johannes Gruen Tobias Bauer Theodor Rüber Thomas Schultz |
| author_facet | Johannes Gruen Tobias Bauer Theodor Rüber Thomas Schultz |
| author_sort | Johannes Gruen |
| collection | DOAJ |
| description | Deep learning-based tractography implicitly learns anatomical prior knowledge that is required to resolve ambiguities inherent in traditional streamline tractography. TractSeg is a particularly widely used example of such an approach. Even though it has exclusively been trained on healthy subjects, a certain level of generalization to different pathologies has been demonstrated, and TractSeg is now increasingly used for clinical cases. We explore the limits of TractSeg by evaluating it on a unique dataset of 25 patients with epilepsy who underwent hemispherotomy, a type of surgery in which the two hemispheres are surgically separated. We compare results to those on 25 healthy controls who have been imaged with the same setup.We find that TractSeg generalizes remarkably well, given the severity of the abnormalities. However, to our knowledge, we are the first to document cases in which TractSeg erroneously reconstructs (“hallucinates”) tracts that are known to have been surgically disconnected, and we found cases in which it implausibly continues tracts through obvious lesions. At the same time, TractSeg failed to reconstruct or undersegmented some tracts that are known to be preserved.We subsequently propose a refinement of TractSeg which aims to improve its applicability to data with pathologies, by using its Tract Orientation Maps as an anatomical prior in low-rank tensor approximation based tractography such that tracking is guaranteed to continue only where presence of the tract is directly supported by the data (“data fidelity”). We demonstrate that our extension not only eliminates hallucinated tracts and reconstructions within lesions, but that it also increases the ability to reconstruct the preserved tracts, and leads to more complete reconstructions even in healthy controls. Despite these advances, we recommend caution and manual quality control when applying deep learning based tractography to patient data. |
| format | Article |
| id | doaj-art-c397e3f5f84340aca071be044ce12077 |
| institution | Kabale University |
| issn | 2213-1582 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Elsevier |
| record_format | Article |
| series | NeuroImage: Clinical |
| spelling | doaj-art-c397e3f5f84340aca071be044ce120772025-02-08T05:00:22ZengElsevierNeuroImage: Clinical2213-15822025-01-0145103738Deep learning based tractography with TractSeg in patients with hemispherotomy: Evaluation and refinementJohannes Gruen0Tobias Bauer1Theodor Rüber2Thomas Schultz3B-IT and Institute for Computer Science, University of Bonn, Friedrich-Hirzebruch-Allee 8, Bonn, 53115, Germany; Corresponding author.Department of Neuroradiology, University Hospital Bonn, Venusberg-Campus 1, Bonn, 53127, Germany; Department of Epileptology, University Hospital Bonn, Venusberg-Campus 1, Bonn, 53127, Germany; German Center for Neurodegenerative Diseases (DZNE), Venusberg-Campus 1, Bonn, 53127, GermanyDepartment of Neuroradiology, University Hospital Bonn, Venusberg-Campus 1, Bonn, 53127, Germany; Department of Epileptology, University Hospital Bonn, Venusberg-Campus 1, Bonn, 53127, Germany; German Center for Neurodegenerative Diseases (DZNE), Venusberg-Campus 1, Bonn, 53127, Germany; Center for Medical Data Usability and Translation, University of Bonn, Regina-Pacis-Weg 3, Bonn, 53113, GermanyB-IT and Institute for Computer Science, University of Bonn, Friedrich-Hirzebruch-Allee 8, Bonn, 53115, Germany; Lamarr Institute for Machine Learning and Artificial Intelligence, Germany 11 https://lamarr-institute.org/.Deep learning-based tractography implicitly learns anatomical prior knowledge that is required to resolve ambiguities inherent in traditional streamline tractography. TractSeg is a particularly widely used example of such an approach. Even though it has exclusively been trained on healthy subjects, a certain level of generalization to different pathologies has been demonstrated, and TractSeg is now increasingly used for clinical cases. We explore the limits of TractSeg by evaluating it on a unique dataset of 25 patients with epilepsy who underwent hemispherotomy, a type of surgery in which the two hemispheres are surgically separated. We compare results to those on 25 healthy controls who have been imaged with the same setup.We find that TractSeg generalizes remarkably well, given the severity of the abnormalities. However, to our knowledge, we are the first to document cases in which TractSeg erroneously reconstructs (“hallucinates”) tracts that are known to have been surgically disconnected, and we found cases in which it implausibly continues tracts through obvious lesions. At the same time, TractSeg failed to reconstruct or undersegmented some tracts that are known to be preserved.We subsequently propose a refinement of TractSeg which aims to improve its applicability to data with pathologies, by using its Tract Orientation Maps as an anatomical prior in low-rank tensor approximation based tractography such that tracking is guaranteed to continue only where presence of the tract is directly supported by the data (“data fidelity”). We demonstrate that our extension not only eliminates hallucinated tracts and reconstructions within lesions, but that it also increases the ability to reconstruct the preserved tracts, and leads to more complete reconstructions even in healthy controls. Despite these advances, we recommend caution and manual quality control when applying deep learning based tractography to patient data.http://www.sciencedirect.com/science/article/pii/S2213158225000087TractographyDeep learningData fidelity |
| spellingShingle | Johannes Gruen Tobias Bauer Theodor Rüber Thomas Schultz Deep learning based tractography with TractSeg in patients with hemispherotomy: Evaluation and refinement NeuroImage: Clinical Tractography Deep learning Data fidelity |
| title | Deep learning based tractography with TractSeg in patients with hemispherotomy: Evaluation and refinement |
| title_full | Deep learning based tractography with TractSeg in patients with hemispherotomy: Evaluation and refinement |
| title_fullStr | Deep learning based tractography with TractSeg in patients with hemispherotomy: Evaluation and refinement |
| title_full_unstemmed | Deep learning based tractography with TractSeg in patients with hemispherotomy: Evaluation and refinement |
| title_short | Deep learning based tractography with TractSeg in patients with hemispherotomy: Evaluation and refinement |
| title_sort | deep learning based tractography with tractseg in patients with hemispherotomy evaluation and refinement |
| topic | Tractography Deep learning Data fidelity |
| url | http://www.sciencedirect.com/science/article/pii/S2213158225000087 |
| work_keys_str_mv | AT johannesgruen deeplearningbasedtractographywithtractseginpatientswithhemispherotomyevaluationandrefinement AT tobiasbauer deeplearningbasedtractographywithtractseginpatientswithhemispherotomyevaluationandrefinement AT theodorruber deeplearningbasedtractographywithtractseginpatientswithhemispherotomyevaluationandrefinement AT thomasschultz deeplearningbasedtractographywithtractseginpatientswithhemispherotomyevaluationandrefinement |