Application of simultaneous uncertainty quantification and segmentation for oropharyngeal cancer use-case with Bayesian deep learning

Jaakko Sahlsten; Joel Jaskari; Kareem A. Wahid; Sara Ahmed; Enrico Glerean; Renjie He; Benjamin H. Kann; Antti Mäkitie; Clifton D.  Fuller; Mohamed A. Naser; Kimmo Kaski

doi:10.1038/s43856-024-00528-5

Application of simultaneous uncertainty quantification and segmentation for oropharyngeal cancer use-case with Bayesian deep learning

Jaakko Sahlsten, Joel Jaskari, Kareem A. Wahid, Sara Ahmed, Enrico Glerean, Renjie He, Benjamin H. Kann, Antti Mäkitie, Clifton D. Fuller, Mohamed A. Naser, Kimmo Kaski

Tutkimustuotos: Lehtiartikkeli › Article › Scientific › vertaisarvioitu

1 Sitaatiot (Scopus)

7 Lataukset (Pure)

Abstrakti

Background

Radiotherapy is a core treatment modality for oropharyngeal cancer (OPC), where the primary gross tumor volume (GTVp) is manually segmented with high interobserver variability. This calls for reliable and trustworthy automated tools in clinician workflow. Therefore, accurate uncertainty quantification and its downstream utilization is critical.

Methods

Here we propose uncertainty-aware deep learning for OPC GTVp segmentation, and illustrate the utility of uncertainty in multiple applications. We examine two Bayesian deep learning (BDL) models and eight uncertainty measures, and utilize a large multi-institute dataset of 292 PET/CT scans to systematically analyze our approach.

Results

We show that our uncertainty-based approach accurately predicts the quality of the deep learning segmentation in 86.6% of cases, identifies low performance cases for semi-automated correction, and visualizes regions of the scans where the segmentations likely fail.

Conclusions

Our BDL-based analysis provides a first-step towards more widespread implementation of uncertainty quantification in OPC GTVp segmentation.

Alkuperäiskieli	Englanti
Artikkeli	110
Sivut	1-12
Sivumäärä	12
Julkaisu	Communications Medicine
Vuosikerta	4
Numero	1
DOI - pysyväislinkit	https://doi.org/10.1038/s43856-024-00528-5
Tila	Julkaistu - jouluk. 2024
OKM-julkaisutyyppi	A1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä

YK:n kestävän kehityksen tavoitteet

Tämä tuotos edistää seuraavia kestävän kehityksen tavoitteita:

Pääsy asiakirjaan

10.1038/s43856-024-00528-5Lisenssi: CC BY

Application of simultaneous uncertainty quantification and segmentation for oropharyngeal cancer use-case with Bayesian deep learningFinal published version, 1,72 MBLisenssi: CC BY

Muut tiedostot ja linkit

XAIS/Lampinen: eXplainable AI Technologies for Segmenting 3D Imaging Data
Lampinen, J. (Vastuullinen tutkija), Jaskari, J. (Projektin jäsen), Sahlsten, J. (Projektin jäsen), Kaski, K. (Projektin jäsen), Pykälä, L. (Projektin jäsen), Saukkoriipi, M. (Projektin jäsen), Takko, T. (Projektin jäsen) & Pienimäki, P. (Projektin jäsen)
01/01/2022 → 31/12/2024
Projekti: Academy of Finland: Other research funding

Siteeraa tätä

Sahlsten, J., Jaskari, J., Wahid, K. A., Ahmed, S., Glerean, E., He, R., Kann, B. H., Mäkitie, A., Fuller, C. D., Naser, M. A., & Kaski, K. (2024). Application of simultaneous uncertainty quantification and segmentation for oropharyngeal cancer use-case with Bayesian deep learning. Communications Medicine, 4(1), 1-12. Artikkeli 110. https://doi.org/10.1038/s43856-024-00528-5

@article{97f42e7d07dd4afcbb660edc48069de7,

title = "Application of simultaneous uncertainty quantification and segmentation for oropharyngeal cancer use-case with Bayesian deep learning",

abstract = "BackgroundRadiotherapy is a core treatment modality for oropharyngeal cancer (OPC), where the primary gross tumor volume (GTVp) is manually segmented with high interobserver variability. This calls for reliable and trustworthy automated tools in clinician workflow. Therefore, accurate uncertainty quantification and its downstream utilization is critical.MethodsHere we propose uncertainty-aware deep learning for OPC GTVp segmentation, and illustrate the utility of uncertainty in multiple applications. We examine two Bayesian deep learning (BDL) models and eight uncertainty measures, and utilize a large multi-institute dataset of 292 PET/CT scans to systematically analyze our approach.ResultsWe show that our uncertainty-based approach accurately predicts the quality of the deep learning segmentation in 86.6% of cases, identifies low performance cases for semi-automated correction, and visualizes regions of the scans where the segmentations likely fail.ConclusionsOur BDL-based analysis provides a first-step towards more widespread implementation of uncertainty quantification in OPC GTVp segmentation.",

author = "Jaakko Sahlsten and Joel Jaskari and Wahid, {Kareem A.} and Sara Ahmed and Enrico Glerean and Renjie He and Kann, {Benjamin H.} and Antti M{\"a}kitie and Fuller, {Clifton D.} and Naser, {Mohamed A.} and Kimmo Kaski",

note = "The work of Joel Jaskari, Jaakko Sahlsten, and Kimmo K. Kaski was supported in part by the Academy of Finland under Project 345449. Antti M{\"a}kitie is supported in part by a grant from the Finnish Society of Sciences and Letters. Kareem A. Wahid is supported by the Dr. John J. Kopchick Fellowship through The University of Texas MD Anderson UTHealth Graduate School of Biomedical Sciences, the American Legion Auxiliary Fellowship in Cancer Research, and an NIH/National Institute for Dental and Craniofacial Research (NIDCR) F31 fellowship (1 F31DE031502-01) and the NCI NRSA Image Guided Cancer Therapy Training Program (T32CA261856). Benjamin H. Kann is supported by an NIH/National Institute for Dental and Craniofacial Research (NIDCR) K08 Grant (K08DE030216). Clifton D. Fuller receives related grant support from the NCI NRSA Image Guided Cancer Therapy Training Program (T32CA261856), as well as additional unrelated salary/effort support from NIH institutes. Dr. Fuller receives grant and infrastructure support from MD Anderson Cancer Center via: the Charles and Daneen Stiefel Center for Head and Neck Cancer Oropharyngeal Cancer Research Program; the Program in Image-guided Cancer Therapy; and the NIH/NCI Cancer Center Support Grant (CCSG) Radiation Oncology and Cancer Imaging Program (P30CA016672). Dr. Fuller has received unrelated direct industry grant/in-kind support, honoraria, and travel funding from Elekta AB.",

year = "2024",

month = dec,

doi = "10.1038/s43856-024-00528-5",

language = "English",

volume = "4",

pages = "1--12",

journal = "Communications Medicine",

issn = "2730-664X",

publisher = "Springer",

number = "1",

}

TY - JOUR

T1 - Application of simultaneous uncertainty quantification and segmentation for oropharyngeal cancer use-case with Bayesian deep learning

AU - Sahlsten, Jaakko

AU - Jaskari, Joel

AU - Wahid, Kareem A.

AU - Ahmed, Sara

AU - Glerean, Enrico

AU - He, Renjie

AU - Kann, Benjamin H.

AU - Mäkitie, Antti

AU - Fuller, Clifton D.

AU - Naser, Mohamed A.

AU - Kaski, Kimmo

N1 - The work of Joel Jaskari, Jaakko Sahlsten, and Kimmo K. Kaski was supported in part by the Academy of Finland under Project 345449. Antti Mäkitie is supported in part by a grant from the Finnish Society of Sciences and Letters. Kareem A. Wahid is supported by the Dr. John J. Kopchick Fellowship through The University of Texas MD Anderson UTHealth Graduate School of Biomedical Sciences, the American Legion Auxiliary Fellowship in Cancer Research, and an NIH/National Institute for Dental and Craniofacial Research (NIDCR) F31 fellowship (1 F31DE031502-01) and the NCI NRSA Image Guided Cancer Therapy Training Program (T32CA261856). Benjamin H. Kann is supported by an NIH/National Institute for Dental and Craniofacial Research (NIDCR) K08 Grant (K08DE030216). Clifton D. Fuller receives related grant support from the NCI NRSA Image Guided Cancer Therapy Training Program (T32CA261856), as well as additional unrelated salary/effort support from NIH institutes. Dr. Fuller receives grant and infrastructure support from MD Anderson Cancer Center via: the Charles and Daneen Stiefel Center for Head and Neck Cancer Oropharyngeal Cancer Research Program; the Program in Image-guided Cancer Therapy; and the NIH/NCI Cancer Center Support Grant (CCSG) Radiation Oncology and Cancer Imaging Program (P30CA016672). Dr. Fuller has received unrelated direct industry grant/in-kind support, honoraria, and travel funding from Elekta AB.

PY - 2024/12

Y1 - 2024/12

N2 - BackgroundRadiotherapy is a core treatment modality for oropharyngeal cancer (OPC), where the primary gross tumor volume (GTVp) is manually segmented with high interobserver variability. This calls for reliable and trustworthy automated tools in clinician workflow. Therefore, accurate uncertainty quantification and its downstream utilization is critical.MethodsHere we propose uncertainty-aware deep learning for OPC GTVp segmentation, and illustrate the utility of uncertainty in multiple applications. We examine two Bayesian deep learning (BDL) models and eight uncertainty measures, and utilize a large multi-institute dataset of 292 PET/CT scans to systematically analyze our approach.ResultsWe show that our uncertainty-based approach accurately predicts the quality of the deep learning segmentation in 86.6% of cases, identifies low performance cases for semi-automated correction, and visualizes regions of the scans where the segmentations likely fail.ConclusionsOur BDL-based analysis provides a first-step towards more widespread implementation of uncertainty quantification in OPC GTVp segmentation.

AB - BackgroundRadiotherapy is a core treatment modality for oropharyngeal cancer (OPC), where the primary gross tumor volume (GTVp) is manually segmented with high interobserver variability. This calls for reliable and trustworthy automated tools in clinician workflow. Therefore, accurate uncertainty quantification and its downstream utilization is critical.MethodsHere we propose uncertainty-aware deep learning for OPC GTVp segmentation, and illustrate the utility of uncertainty in multiple applications. We examine two Bayesian deep learning (BDL) models and eight uncertainty measures, and utilize a large multi-institute dataset of 292 PET/CT scans to systematically analyze our approach.ResultsWe show that our uncertainty-based approach accurately predicts the quality of the deep learning segmentation in 86.6% of cases, identifies low performance cases for semi-automated correction, and visualizes regions of the scans where the segmentations likely fail.ConclusionsOur BDL-based analysis provides a first-step towards more widespread implementation of uncertainty quantification in OPC GTVp segmentation.

UR - https://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=aalto_pure&SrcAuth=WosAPI&KeyUT=WOS:001243335000001&DestLinkType=FullRecord&DestApp=WOS_CPL

UR - http://www.scopus.com/inward/record.url?scp=85200794578&partnerID=8YFLogxK

U2 - 10.1038/s43856-024-00528-5

DO - 10.1038/s43856-024-00528-5

M3 - Article

C2 - 38851837

SN - 2730-664X

VL - 4

SP - 1

EP - 12

JO - Communications Medicine

JF - Communications Medicine

IS - 1

M1 - 110

ER -

Application of simultaneous uncertainty quantification and segmentation for oropharyngeal cancer use-case with Bayesian deep learning

Abstrakti

YK:n kestävän kehityksen tavoitteet

Pääsy asiakirjaan

Muut tiedostot ja linkit

Sormenjälki

Projektit

XAIS/Lampinen: eXplainable AI Technologies for Segmenting 3D Imaging Data

Siteeraa tätä