Centered on prostate MRI (T2/DWI/DCE), we build an integrated data platform that combines imaging, radiology reports, and clinical information, and develop deep learning models to improve diagnostic performance and clinical workflow efficiency .

Leveraging Mayo Clinic’s clinical scale, we continuously maintain a large, real-world prostate MRI cohort.

• Building and curating a prostate MRI database since around 2017.
• Integrating MRI (T2/DWI/DCE) with text data such as reports and clinical notes (e.g., PSA, biopsy status, treatment history).
• Validating models on realistic clinical data that include artifacts (e.g., hip arthroplasty-related DWI artifacts) and variable image quality.

• Prostate cancer detection model (MRI, weakly supervised): A patient-level labeling approach using T2/DWI/DCE as inputs, reducing the need for lesion-level annotations.
• Zonal segmentation (U-Net, etc.): Prostate zonal segmentation to derive volumetric biomarkers and evaluate clinical utility such as TZ-PSAD .
• Image quality assessment (QC) models: Automated quality assessment for T2/DWI to reduce unnecessary rescans and mitigate variability; includes bias analyses.
• NLP (clinical notes / reports): Automated extraction of clinically relevant information for interpretation (history/treatment, family history, DRE, biopsy status, etc.) to support decision-making.
• Performance metrics and estimated diagnostic performance: A framework combining CDR/AIR/PPV and methods to estimate sensitivity/specificity/AUC in settings including non-biopsied cases; also supports internal feedback and educational case review.

• External validation to assess generalizability.
• Improving localization performance using ROI / lesion-level labels .
• Preparing clinical deployment of cancer detection models.
• Bias detection and mitigation for QC models (including confounders such as prostate volume).
• Improving performance via multimodal fusion of imaging + report/clinical information.
• Developing prediction models for outcomes beyond cancer presence (e.g., extraprostatic extension ).

Ultimately, our goal is clinical adoption that improves reproducibility and interpretation efficiency.

The numbers correspond to the publication list below. “(PI: Name)” indicates the principal investigator / senior author for that work. Items without a PI label are projects I led.

• #7: Weakly supervised prostate cancer detection model using T2/DWI/DCE with patient-level labels only (no lesion location annotations).
• #9: Prostate zonal segmentation model; derives TZ volume to evaluate TZ-PSAD as a clinical biomarker. Also used as an input component for #7.
• #13: Risk prediction model using report-derived features and clinical variables (e.g., PI-RADS, PSAD); evaluated the contribution of ADC (challenges in external validation).

• #6: Impact of hip arthroplasty-related DWI artifacts; highlights the value of CDR in addition to PPV.
• #12: Same-patient comparison of 1.5T vs 3T in post-hip arthroplasty cases to evaluate differences in DWI artifacts (PI: Kawashima).
• #15: Deep learning QC model for automated T2WI image quality assessment (PI: Riederer).
• #17: Effect of rectal gas–related susceptibility artifacts in DWI; uses a deep learning QC model for quality categorization.
• #19: Bias analysis of the T2WI QC model (e.g., potential confounding such as prostate volume influencing quality predictions).
• #20: Paired assessment showing glucagon can improve T2WI image quality (PI: Riederer).

• #11: NLP model to extract clinically relevant information from notes to support MRI interpretation (history/treatment, family history, DRE, biopsy status, etc.).

• #4–5: Proposes using both CDR (cancer detection rate) and AIR (abnormal interpretation rate) as performance metrics for prostate MRI.
• #10: Estimation of diagnostic performance (sensitivity/specificity/AUC, etc.) in settings that include non-biopsied cases.

• #8: Study on T2WI reconstruction (PI: Riederer).
• #14: Study on prostate MRI coils (PI: Riederer).
• #16: Sequence design aligned with PI-RADS v2.1 in-plane resolution requirements, achieving shorter acquisition time and improved image quality (PI: Riederer).

• #18: Lesion detection/localization using PSMA PET-CT and prostate MRI (e.g., clinical significance of PSMA uptake in PI-RADS 3 lesions) (PI: H. Takahashi).

1. Hassanzadeh, S., Borisch, E. A., Froemming, A. T., Kawashima, A., Takahashi, N., & Riederer, S. J. (in press). Prostate T2-weighted spin-echo MRI with and without glucagon: A paired scan quality assessment. Abdominal Radiology . https://doi.org/10.1007/s00261-025-05215-0
2. Nakai, H., Froemming, A. T., Kawashima, A., Legout, J. D., Kurata, Y., Gloe, J. N., Borisch, E. A. Riederer, S. J., & Takahashi, N. (in press). Bias in deep learning-based image quality assessments of T2-weighted imaging in prostate MRI. Abdominal Radiology . https://doi.org/10.1007/s00261-025-05163-9
3. Takahashi, H., Nakai, H., Ballman, K. V., Lomas, D. J., Mynderse, L. A., Kawashima, A., Huang, S., Legout, J. D., Young, J. R., Thorpe, M. P., Johnson, G. B., Karnes, R. J., Sartor, A. O., & Takahashi, N. (2025). Localization of PSMA-avid lesions on PSMA PET-CT on prostate MRI in patients with PI-RADS 3. Abdominal Radiology , 50 (12), 5948–5962. https://doi.org/10.1007/s00261-025-05018-3
4. Nakai, H., Froemming, A. T., Takahashi, H., Adamo, D. A., Kawashima, A., LeGout, J. D., Kurata, Y., Gloe, J. N., Borisch, E. A., Riederer, S. J., & Takahashi, N. (2025). Prostate MRI cancer detection rate by deep learning-assisted image quality categorization: Gas-induced susceptibility artifacts in diffusion-weighted imaging. Insights into Imaging , 16 (1), 217. https://doi.org/10.1186/s13244-025-02110-6
5. Riederer, S. J., Borisch, E. A., Froemming, A. T., Grimm, R. C., Hassanzadeh, S., Kawashima, A., Takahashi, N., & Thomas, J. (2025). Improved image quality and reduced acquisition time in prostate T2-weighted spin-echo MRI using a modified PI-RADS-adherent sequence. European Radiology Experimental , 9 (1), 55. https://doi.org/10.1186/s41747-025-00595-w
6. Gloe, J. N., Borisch, E. A., Froemming, A. T., Kawashima, A., LeGout, J. D., Nakai, H., Takahashi, N., & Riederer, S. J. (2025). Deep learning for quality assessment of axial T2-weighted prostate MRI: A tool to reduce unnecessary rescanning. European Radiology Experimental , 9 (1), 44. https://doi.org/10.1186/s41747-025-00584-z
7. Riederer, S. J., Borisch, E. A., Du, Q., Froemming, A. T., Hulshizer, T. C., Kawashima, A., McGee, K. P., Robb, F., Rossman, P. J., & Takahashi, N. (2025). Application of high-density 2D receiver coil arrays for improved SNR in prostate MRI. Magnetic Resonance in Medicine , 93 (2), 850–863. https://doi.org/10.1002/mrm.30289
8. Nakai, H., Takahashi, H., LeGout, J. D., Kawashima, A., Froemming, A. T., Klug, J. R., Korfiatis, P., Lomas, D. J., Humphreys, M. R., Dora, C., & Takahashi, N. (2025). Prostate Cancer Risk Prediction Model Using Clinical and Magnetic Resonance Imaging–Related Findings: Impact of Combining Lesions’ Locations and Apparent Diffusion Coefficient Values. Journal of Computer Assisted Tomography , 49 (2), 247–257. https://doi.org/10.1097/RCT.0000000000001679
9. Nakai, H., Takahashi, N., Sugi, M. D., Wellnitz, C. V., Thompson, C. P., & Kawashima, A. (2024). Image quality comparison of 1.5T and 3T prostate MRIs of the same post-hip arthroplasty patients: Multi-rater assessments including PI-QUAL version 2. Abdominal Radiology , 49 (11), 3913–3924. https://doi.org/10.1007/s00261-024-04483-6
10. Nakai, H., Suman, G., Adamo, D. A., Navin, P. J., Bookwalter, C. A., LeGout, J. D., Chen, F. K., Wellnitz, C. V., Silva, A. C., Thomas, J. V., Kawashima, A., Fan, J. W., Froemming, A. T., Lomas, D. J., Humphreys, M. R., Dora, C., Korfiatis, P., & Takahashi, N. (2024). Natural language processing pipeline to extract prostate cancer-related information from clinical notes. European Radiology , 34 (12), 7878–7891. https://doi.org/10.1007/s00330-024-10812-6
11. Nakai, H., Takahashi, H., LeGout, J. D., Kawashima, A., Froemming, A. T., Lomas, D. J., Humphreys, M. R., Dora, C., & Takahashi, N. (2024). Estimated diagnostic performance of prostate MRI performed with clinical suspicion of prostate cancer. Insights into Imaging , 15 (1), 271. https://doi.org/10.1186/s13244-024-01845-y
12. Kuanar, S., Cai, J. C., Nakai, H., Takahashi, H., LeGout, J. D., Kawashima, A., Froemming, A. T., Mynderse, L. A., Dora, C., Humphreys, M. R., Klug, J., Korfiatis, P., Erickson, B., & Takahashi, N. (2024). Transition-Zone PSA-Density Calculated from MRI Deep Learning Prostate Zonal Segmentation Model for Prediction of Clinically Significant Prostate Cancer. Abdominal Radiology , 49 (10), 3722-3734. https://doi.org/10.1007/s00261-024-04301-z
13. Riederer, S. J., Borisch, E. A., Froemming, A. T., Kawashima, A., & Takahashi, N. (2024). Comparison of model-based versus deep learning-based image reconstruction for thin-slice T2-weighted spin-echo prostate MRI. Abdominal Radiology , 49 (8), 2921–2931. https://doi.org/10.1007/s00261-024-04256-1
14. Cai, J. C., Nakai, H., Kuanar, S., Froemming, A. T., Bolan, C. W., Kawashima, A., Takahashi, H., Mynderse, L. A., Dora, C. D., Humphreys, M. R., Korfiatis, P., Rouzrokh, P., Bratt, A. K., Conte, G. M., Erickson, B. J., & Takahashi, N. (2024). Fully Automated Deep Learning Model to Detect Clinically Significant Prostate Cancer at MRI. Radiology , 312 (2), e232635. https://doi.org/10.1148/radiol.232635
15. Nakai, H., Takahashi, H., Adamo, D. A., LeGout, J. D., Kawashima, A., Thomas, J. V., Froemming, A. T., Kuanar, S., Lomas, D. J., Humphreys, M. R., Dora, C., & Takahashi, N. (2024). Decreased prostate MRI cancer detection rate due to moderate to severe susceptibility artifacts from hip prosthesis. European Radiology , 34 (5), 3387-3393. https://doi.org/10.1007/s00330-023-10345-4
16. Nagayama, H., Nakai, H., Takahashi, H., Froemming, A. T., Kawashima, A., Bolan, C. W., Adamo, D. A., Carter, R. E., Fazzio, R. T., Tsuji, S., Lomas, D. J., Mynderse, L. A., Humphreys, M. R., Dora, C., & Takahashi, N. (2024). Cancer Detection Rate and Abnormal Interpretation Rate of Prostate MRI Performed for Clinical Suspicion of Prostate Cancer. Journal of the American College of Radiology , 21 (3), 398–408. https://doi.org/10.1016/j.jacr.2023.07.031
17. Nakai, H., Nagayama, H., Takahashi, H., Froemming, A. T., Kawashima, A., Bolan, C. W., Adamo, D. A., Carter, R. E., Fazzio, R. T., Tsuji, S., Lomas, D. J., Mynderse, L. A., Humphreys, M. R., Dora, C., & Takahashi, N. (2024). Cancer Detection Rate and Abnormal Interpretation Rate of Prostate MRI in Patients With Low-Grade Cancer. Journal of the American College of Radiology , 21 (3), 387–397. https://doi.org/10.1016/j.jacr.2023.07.030
18. Takahashi, H., Yoshida, K., Kawashima, A., Lee, N. J., Froemming, A. T., Adamo, D. A., Khandelwal, A., Bolan, C. W., Heller, M. T., Hartman, R. P., Kim, B., Philbrick, K. A., Carter, R. E., Mynderse, L. A., Humphreys, M. R., Cai, J. C., & Takahashi, N. (2022). Impact of measurement method on interobserver variability of apparent diffusion coefficient of lesions in prostate MRI. PLOS ONE , 17 (5), e0268829. https://doi.org/10.1371/journal.pone.0268829
19. Takahashi, H., Froemming, A. T., Bruining, D. H., Karnes, R. J., Jimenez, R. E., & Takahashi, N. (2021). Prostate MRI characteristics in patients with inflammatory bowel disease. European Journal of Radiology , 135 , 109503. https://doi.org/10.1016/j.ejrad.2020.109503
20. Yoshida, K., Takahashi, N., Karnes, R. J., & Froemming, A. T. (2020). Prostatic Remnant After Prostatectomy: MR Findings and Prevalence in Clinical Practice. American Journal of Roentgenology , 214 (1), W37–W43. https://doi.org/10.2214/AJR.19.21345