EUVIMEDEUVIMED entdecken
Crossref

Balancing Bias and Variance in Deep Learning‐Based Tumor Microstructural Parameter Mapping

Jiaren Zou iD, Yue Cao

Magnetic Resonance in Medicine · 2026

DOI: 10.1002/mrm.70154 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Magnetic Resonance in Medicine, 2026. Verfasst von Jiaren Zou und 1 weiteren Autor:innen. Beteiligte Einrichtung: Department of Radiation Oncology, University of Michigan, Ann Arbor, Michigan, USA. Hauptthemen laut MEDLINE: Deep Learning, Head and Neck Neoplasms, Diffusion Magnetic Resonance Imaging, Image Processing, Computer-Assisted. Die Arbeit nennt 36 Literaturangaben. Gefördert von: NIH HHS, National Institutes of Health.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Purpose

Time‐dependent diffusion MRI enables quantification of tumor microstructural parameters useful for diagnosis and prognosis. Nevertheless, current model fitting approaches exhibit suboptimal bias‐variance trade‐offs; specifically, nonlinear least squares fitting (NLLS) demonstrated low bias but high variance, whereas supervised deep learning methods trained with mean squared error loss (MSE‐Net) yielded low variance but elevated bias. This study investigates these bias‐variance characteristics and proposes a method to control fitting bias and variance.

Methods

Random walk with barrier model was used as a representative biophysical model. NLLS and MSE‐Net were reformulated within the Bayesian framework to elucidate their bias‐variance behaviors. We introduced B2V‐Net, a supervised learning approach using a loss function with adjustable bias‐variance weighting, to control bias‐variance trade‐off. B2V‐Net was evaluated and compared against NLLS and MSE‐Net numerically across a wide range of parameters and noise levels, as well as in vivo in patients with head and neck cancer.

Results

Flat posterior distributions that were not centered at ground truth parameters explained the bias‐variance behaviors of NLLS and MSE‐Net. B2V‐Net controlled the bias‐variance trade‐off, achieving a 56% reduction in standard deviation relative to NLLS and an 18% reduction in bias compared to MSE‐Net. In vivo parameter maps from B2V‐Net demonstrated a balance between smoothness and accuracy.

Conclusion

We demonstrated and explained the low bias‐high variance of NLLS and the low variance‐high bias of MSE‐Net. The proposed B2V‐Net can balance bias and variance. Our work provided insights and methods to guide the design of customized loss functions tailored to specific clinical imaging needs.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningHead and Neck NeoplasmsDiffusion Magnetic Resonance ImagingImage Processing, Computer-AssistedImage Interpretation, Computer-AssistedHumansAlgorithmsBayes TheoremLeast-Squares AnalysisBias

Stichwörter der Autor:innen

deep learningdiffusion MRIhead and neck cancersmodel fittingtissue microstructure

Autor:innen und Institutionen

  • Department of Radiation Oncology, University of Michigan, Ann Arbor, Michigan, USA.
  • Department of Radiology, University of Michigan, Ann Arbor, Michigan, USA.
  • Department of Biomedical Engineering, University of Michigan, Ann Arbor, Michigan, USA.

ORCID-Profile: Jiaren Zou

Förderung

  • NIH HHS
  • National Institutes of Health

Zitieren

APA

Zou, J., & Cao, Y. (2026). Balancing Bias and Variance in Deep Learning‐Based Tumor Microstructural Parameter Mapping. Magnetic Resonance in Medicine. https://doi.org/10.1002/mrm.70154

Vancouver

Zou J, Cao Y. Balancing Bias and Variance in Deep Learning‐Based Tumor Microstructural Parameter Mapping. Magnetic Resonance in Medicine. 2026. doi:10.1002/mrm.70154

BibTeX
@article{Zou2026,
  author = {Zou, Jiaren and Cao, Yue},
  title = {Balancing Bias and Variance in Deep Learning‐Based Tumor Microstructural Parameter Mapping},
  journal = {Magnetic Resonance in Medicine},
  year = {2026},
  doi = {10.1002/mrm.70154}
}
Kostenlos starten →