Scientific job: CLAIMED. This article owns the public treatment-delivery movement from radiotherapy intent → simulation and immobilisation → target and organs-at-risk definition → dose prescription → treatment planning and dosimetry → physics quality assurance → image verification → fraction delivery → treatment-course completion → toxicity and response return. The Oncology Web retains ownership of the cancer trajectory, treatment sequencing and overall outcome. Physical World Science retains radiation physics. This node owns how a radiation prescription becomes a reproducible treatment delivered to a real patient.

Wait, what? Radiotherapy is not “aim radiation at the tumour”.

Radiotherapy is a chain of measurements, representations, calculations and repeated verification. The target can move with breathing, filling of an organ can change anatomy, a patient can lose weight during treatment, and the same physical radiation dose can affect tumour and normal tissues differently depending on where it lands and how it is fractionated.

The important eduKateAI correction is therefore: radiation prescribed ≠ radiation safely delivered. The intended treatment has to survive every handoff from oncologist to simulation, contouring, planning, medical physics, treatment machine, image guidance and repeated fractions.

The radiotherapy tube

Cancer state + treatment intent → radiation-oncology decision → simulation → immobilisation → target and organs-at-risk contours → dose/fraction prescription → planning optimisation → dosimetric review → physics QA → patient identity + plan verification → image-guided setup → fraction delivery → repeat verification → toxicity/support → completion → response and survivorship handoff.

1. Treatment intent comes before the machine

Radiotherapy can be used with curative, adjuvant, neoadjuvant, consolidative, disease-control or symptom-relieving intent depending on cancer type and clinical context. The same anatomical site can therefore receive very different treatment plans because the purpose is different.

For eduKateAI, the radiation object must preserve cancer diagnosis, anatomical site, stage or disease extent where relevant, treatment intent, systemic/surgical sequencing, prior radiation exposure and the clinical outcome being sought.

2. Simulation creates the coordinate system

Before many forms of external-beam radiotherapy, a planning CT or other simulation process is used to reproduce the patient’s treatment position and capture anatomy for planning. Immobilisation devices can help the patient return to the same position each day.

Simulation is not merely another diagnostic scan. It is a treatment geometry object. Position, supports, masks, marks, breathing instructions and preparation state can all become part of the future delivery contract.

3. Target definition is not identical to the visible tumour

Radiation oncologists may define visible tumour, areas at risk of microscopic disease and planning margins that account for movement and setup uncertainty. Normal tissues that should be protected are separately delineated as organs at risk.

This creates a clean anti-collapse rule: image abnormality ≠ radiation target automatically. Pathology, imaging, operative findings, anatomy and disease behaviour can all contribute to target definition.

4. Dose and fractionation are two parts of one prescription

Radiotherapy is often divided into multiple treatment sessions called fractions. Total dose, dose per fraction, number of fractions and treatment schedule interact biologically. A treatment prescribed over one, five or thirty fractions is not the same clinical object even if the same anatomical region is involved.

For eduKateAI: total dose ≠ whole prescription. The route should preserve total dose, fraction size, number of fractions, treatment frequency and any boost or adaptive phase.

5. Treatment planning is an optimisation problem

Planning attempts to achieve adequate dose coverage of the intended target while reducing unnecessary dose to normal tissues. Modern systems may use multiple beam angles, intensity modulation, arcs, stereotactic techniques or other delivery strategies depending on the case.

The planning question is therefore not “Can radiation reach this site?” but Can the intended target receive an acceptable distribution while respecting clinically relevant normal-tissue constraints?

6. Medical Physics is a safety owner, not a background service

Medical physicists help verify dose calculation, treatment-machine performance, plan deliverability and patient-specific quality assurance where indicated. NCCS describes medical physics as part of the radiotherapy team responsible for treatment planning, equipment calibration, quality assurance and safe radiation delivery.

For eduKateAI, the plan should not move from software directly to patient without a visible physics verification state.

7. Image guidance turns every fraction into a new positioning question

Even with good immobilisation, day-to-day anatomy can vary. Image-guided radiotherapy can compare current patient anatomy with the planned geometry before treatment, allowing selected corrections in position or strategy.

For eduKateAI: correct plan ≠ correct setup today. Fraction-level verification is a separate receipt.

8. Fraction delivery is repetitive but never trivial

A course may involve many repeated visits. Identity, correct plan, correct site, current patient state and treatment parameters must remain reliable every time. The repetition creates opportunities for both safety and error.

The useful runtime is identify → verify anatomy → deliver → record → observe → repeat, not “set once and forget”.

9. Anatomy can change during treatment

Weight loss, tumour shrinkage, oedema, organ filling or other changes can make the original geometry less representative later in the course. Selected patients may therefore need review, repeat imaging or adaptive replanning.

For eduKateAI: plan approved once ≠ plan necessarily valid forever. A replan should preserve why it occurred and how the new plan relates to the previous treatment already delivered.

10. Toxicity is part of the radiation state

Acute effects can arise during or shortly after treatment; late effects can emerge months or years later. The pattern depends on site, dose, fractionation, normal-tissue exposure, concurrent therapies and individual susceptibility.

Oncology integrates the overall cancer-treatment toxicity. Radiotherapy owns the link between irradiated volume + dose history + timing and radiation-specific effects.

11. Treatment interruption is a clinical state

A missed fraction can occur because of illness, machine downtime, toxicity, transport or other barriers. The clinical significance depends on disease, fractionation and duration of interruption. The system should therefore track missed treatment as a meaningful event rather than a scheduling footnote.

For eduKateAI, preserve planned date, delivered date, reason for interruption and authorised compensatory plan if any.

12. Completion is not the endpoint

After the last fraction, patients may need response imaging, toxicity review, symptom management, rehabilitation and long-term surveillance for recurrence or late effects. The Oncology Web owns the disease trajectory; Primary Care and survivorship pathways may become long-term receivers.

The radiation receipt is therefore: planned course delivered as intended, deviations explained, acute effects managed, response/late-effect follow-up accepted by the next owner.

Characteristic failure modes

• Tumour = target error: visible lesion treated as the whole radiation target without disease-context reasoning.
• Total dose = prescription error: fraction size and schedule disappear.
• Plan = delivery error: approved plan treated as proof that treatment reached the intended anatomy.
• Physics-invisible error: dose calculation and machine QA disappear from the clinical chain.
• First-day = every-day error: changing anatomy and setup are ignored across fractions.
• Completion = cure error: course completion treated as final cancer outcome.
• Late-effects loss: long-term toxicity is detached from the original radiation history.

The eduKateAI routing contract

• Canonical public owner: Radiation Oncology & Radiotherapy Web.
• Input state: patient with a radiation-oncology treatment decision.
• Primary job: preserve intent, simulation geometry, target/OAR definition, dose/fraction prescription, plan, QA, daily verification, delivered fractions and toxicity as one auditable treatment lineage.
• Do not collapse: cancer diagnosis ≠ radiation target; total dose ≠ full prescription; approved plan ≠ delivered dose; treatment completed ≠ disease outcome.
• Handoffs: Oncology, Radiology, Anatomical Pathology, Medical Physics, Surgery, Clinical Nutrition, Rehabilitation, Palliative Care and relevant organ specialties.
• Return receipt: simulation complete, plan approved, QA passed, fractions delivered/missed, toxicity acceptable/unacceptable, course completed, response/survivorship owner confirmed.

Authoritative routes

• National Cancer Centre Singapore — Radiation Therapy
• IAEA — Radiotherapy
• WHO — Radiotherapy as a cancer-care component

Educational boundary: this article explains radiotherapy information architecture. It does not determine whether a patient should receive radiation, define a target, prescribe dose or fractionation, interpret a treatment plan or advise on toxicity for an individual.