EST. 2026 · TOKYO

Institute of One
-Medical Imaging AI Research

Institute of One
-Medical Imaging AI Research

Open, reproducible research in medical imaging physics — every paper ships its code.

Open, reproducible research in medical imaging physics — every paper ships its code.

One Human. A Legion of Agents.

One Human. A Legion of Agents.

A research institute of one.

A research institute of one.

One human directs a legion of AI agents to form a working medical-imaging institute — no campus, no department, no borders.

One human directs a legion of AI agents to form a working medical-imaging institute — no campus, no department, no borders.

We publish the process, not only the results.

We publish the process, not only the results.

The prompts, the decisions, the dead ends, the moments a human overrides the machine — all of it is the record. The process is the contribution.

The prompts, the decisions, the dead ends, the moments a human overrides the machine — all of it is the record. The process is the contribution.

Human-accountable. Agent-accelerated.

Human-accountable. Agent-accelerated.

Every output carries one name and one responsibility: a human author. The agents are credited as tools, in the open. Rigor is never traded for speed.

Every output carries one name and one responsibility: a human author. The agents are credited as tools, in the open. Rigor is never traded for speed.

Built in the open, from anywhere.

Built in the open, from anywhere.

Every result is published with the code and the archived data that produced it, under an open licence, so that anyone can check it.

Shuji Yamamoto, PhD — Founder

Medical imaging researcher (PhD, Health Sciences, Osaka University). Three decades across medical imaging, radiomics, and AI-assisted analysis. Visiting researcher, National Cancer Center Japan · Lecturer in medical imaging & AI, Kochi University School of Medicine. RSNA Certificate of Merit; US and Japanese patents. Accountable for everything published here.

Institute of One, LISIT Co., Ltd., Tokyo 150-0044, Japan

ORCID

Google Scholar

ResearchGate

LISIT CO., LTD. — RESEARCH DIVISION: INSTITUTE OF ONE

METHOD

Research here is carried out by one human author working with AI language models as tools. The models draft, compute, and check; the author sets every hypothesis, verifies every result, and takes sole responsibility for what is published. Model use is disclosed in each manuscript, naming the model version, in line with ICMJE and COPE guidance. AI systems are not authors.

THE LIBRARY

Work is published openly as it happens — Research Notes (IORN), code, and process logs, with DOIs assigned to citable outputs. Everything produced here is a live record of what one person and a legion of agents can build, verify, and share.

Work is published openly as it happens — Research Notes (IORN), code, and process logs, with DOIs assigned to citable outputs. Everything produced here is a live record of what one person and a legion of agents can build, verify, and share.

IORN-001

An Open, Reproducible Gamma-Variate Pipeline for CT-Perfusion Curve Analysis

CODE：https://github.com/Institute-of-One/ctp-core

DOI：https://doi.org/10.5281/zenodo.20921268

PREPRINT：https://www.medrxiv.org/content/10.64898/2026.06.26.26356666v1

IORN-002

A Stability Atlas for IBSI Radiomics Features Using Synthetic Digital Phantoms, with Proof-of-Concept Physics-Based Normalisation

PAPER：https://doi.org/10.3390/jimaging12080392

Journal of Imaging 12(8), 392 (2026) · CC BY 4.0

CODE：https://github.com/Institute-of-One/radiomics-phantom

DOI：https://doi.org/10.5281/zenodo.21309874

IORN-003

An Open, Closed-Form-Validated Framework for Task-Based Image Quality on Synthetic Phantoms

CODE：https://github.com/Institute-of-One/taskiq-core

DOI：https://doi.org/10.5281/zenodo.21422923

STATUS：under review — Journal of Imaging

IORN-004

An Open-Source, Provenance-Aware Framework for Reconstructing Missing CT Dose Indices from DICOM Metadata

CODE：https://github.com/Institute-of-One/ctdose-core

DOI：https://doi.org/10.5281/zenodo.21636082

STATUS：revised Technical Note in preparation — Journal of Applied Clinical Medical Physics

IORN-005

Denoising under a Data-Processing Ceiling: Observer-Dependent Benefits, Fidelity–Task Divergence, and an Information Floor

CODE：https://github.com/Institute-of-One/denoiq-core

DOI：https://doi.org/10.5281/zenodo.21733388

STATUS：under review — Medical Physics

IORN-006

Anatomy-Weighted CTDIvol from Routine CT Metadata: A Patient-Specific, Multi-Vendor Study Using Deep-Learning Segmentation

CODE：https://github.com/Institute-of-One/ctsegdose-core

DOI：https://doi.org/10.5281/zenodo.21817234

STATUS：accepted 28 August 2026 — Tomography

IORN-007

A Measured Floor on Index-Based Organ Dose Estimation in CT, Where Per-Organ Coefficients Do Not Improve on a Single Conversion Factor

CODE：https://github.com/Institute-of-One/ct-dosemc-core

STATUS：manuscript complete

IORN-008

Sampling Design Bounds Parameter Recovery in a Reduced CT Contrast-Kinetics Model, and a Closed-Form Fit Nearly Attains the Bound

CODE：https://github.com/Institute-of-One/sim-ce-core

DOI：https://doi.org/10.5281/zenodo.22037562

STATUS：under review — Computer Methods and Programs in Biomedicine

IORN-009

Where Does Imaging Resolution Stop Reaching the Reader? A Closed-Form Physics-to-Perception Transfer Model with Saturation Frequencies

CODE：https://github.com/Institute-of-One/human_ai_taskcore

DOI：https://doi.org/10.5281/zenodo.22144839

STATUS：manuscript complete

Institute of One is the research division of LISIT Co., Ltd., Tokyo, Japan · instituteofone.org