Oncology Staging & Response CT

What it does

Oncology Staging & Response CT reads the cancer patient's CT the way a tumour board needs it: as one time point in a series. Upload a staging or follow-up CT, usually chest, abdomen and pelvis, together with the earlier studies; the run builds a whole-body lesion inventory, matches each lesion across time points and measures it consistently.

From the inventory it proposes RECIST 1.1 target lesions (at most five, at most two per organ), measures long axes and node short axes, sums the diameters and lists non-target disease. With priors it tracks every lesion, reports new lesions and returns the overall response: complete response, partial response, stable disease or progressive disease, with iRECIST for patients on immunotherapy.

Intended use

Decision support for the radiologist, oncologist and trial team who must produce a reproducible response assessment: the same lesions, measured the same way, at every visit.

Who it is for

Oncologic radiologists, medical and radiation oncologists, residents preparing tumour boards, clinical trial imaging teams and core labs. Developers use the run to fill RECIST case report forms, tumour-board slides or a lesion-tracking viewer.

Accepted input

• Current CT of any region, typically chest-abdomen-pelvis.
• Prior CT studies in prior_files: the baseline and any intermediate follow-ups, so the nadir can be found.
• Contrast-enhanced studies in a comparable phase give the most reliable comparison.

Requirements

Requirement	Why
Same anatomical coverage at each time point	A lesion outside one scan cannot be tracked
Pixel Spacing and Frame of Reference UID	Lesions are measured in millimetres and located in patient coordinates
A baseline study for response	Without priors the run returns staging measurements, not a response category

Different contrast phases between time points, confluent lesions and nodes, and non-measurable disease are flagged on the lesions they affect.

Optional context

Tumour type, treatment line, whether the patient receives immunotherapy, the baseline date and previously chosen target lesions go in clinical_context. Immunotherapy switches the response logic to iRECIST, which handles pseudo-progression.

The result

A sectioned JSON result in which every lesion has a tracking identifier that stays the same across studies.

Section	Content
Lesion inventory	Whole-body lesion and metastasis detection and segmentation with organ attribution and total tumour burden
Lymph nodes	Node detection and segmentation, station, short axis and Node-RADS score
RECIST measurements	Target lesion proposal, long axis (short axis for nodes), sum of diameters and the non-target list
Response against priors	Lesion matching, per-lesion change, new lesions and overall response (CR, PR, SD, PD)
Draft report	Opt-in (options.report): an English narrative that interprets the result for the reader (findings, impression, limitations); every number is checked against the findings

Standards

• Response criteria: RECIST 1.1 for target, non-target and new lesions; iRECIST for immunotherapy.
• Nodes: Node-RADS 1.0.
• DICOM: SR TID 1500 Measurement Report with a tracking identifier and UID per lesion, so a viewer can follow the same lesion from study to study; SEG for lesion masks; key images per target lesion; Encapsulated PDF and FHIR DiagnosticReport.

Response thresholds

Complete response is the disappearance of all target lesions, with every pathological node below 10 mm short axis. Partial response is a decrease of at least 30 % in the sum of diameters from baseline. Progression is an increase of at least 20 % and at least 5 mm from the smallest sum on study (the nadir), a new lesion, or unequivocal progression of non-target disease.