Medical Image
The DICOM-formatted radiology images (X-ray, CT, MRI, ultrasound) with associated metadata including patient context, prior imaging, and clinical indication.
Why This Object Matters for AI
AI imaging analysis models cannot detect abnormalities without standardized image data and clinical context; without it, AI operates blind to patient history affecting interpretation.
Clinical Operations & Patient Care Capacity Profile
Typical CMC levels for clinical operations & patient care in Healthcare organizations.
CMC Dimension Scenarios
What each CMC level looks like specifically for Medical Image. Baseline level is highlighted.
Medical images are taken but not formally stored or documented. An ultrasound image exists on the portable machine's hard drive, an X-ray is printed on film and clipped to the patient's chart, and a wound photo lives on the nurse's personal phone. When the radiologist needs to compare to a prior study, there is no prior study to find.
None — AI imaging analysis cannot function because medical images are not stored in any accessible system and have no associated metadata or clinical context.
Implement a basic PACS (Picture Archiving and Communication System) where all medical images are stored digitally in DICOM format with patient identification and study type metadata.
Medical images are stored in a PACS but with inconsistent metadata. Some studies have complete patient demographics and clinical indication; others have only a patient name and date. Modality worklists are partially used — some technologists enter the correct study type, others leave it blank. Finding a prior comparison study means scrolling through a list of poorly labeled entries.
AI could attempt image analysis on individual studies, but cannot reliably match prior studies for comparison, identify the clinical question being asked, or filter by anatomy or pathology because metadata is inconsistent.
Enforce DICOM metadata standards — require complete modality worklist information (patient ID, accession number, study description, clinical indication, referring provider) for every imaging study before it is stored in PACS.
Medical images are stored in PACS with standardized DICOM metadata. Every study has a patient MRN, accession number, study description, body part examined, clinical indication, and referring provider. The radiology information system (RIS) links orders to completed studies. Comparing to prior studies is straightforward because the metadata is consistent and searchable.
AI can retrieve and organize medical images by patient, body part, and study type for analysis. Basic AI triage (flagging critical findings like pneumothorax or stroke) is possible because images can be correctly identified and routed. Cannot correlate imaging findings with clinical context beyond what is in the DICOM header.
Integrate medical images with the clinical EHR — link imaging studies to encounters, orders, and clinical notes so that AI can access both the image and the clinical context that prompted the study.
Medical images are fully integrated with the clinical record. Each study links to the ordering encounter, the clinical indication, the radiologist's report, and the patient's relevant problem list entries. A clinician can navigate from a chest CT directly to the pulmonary nodule on the problem list, the prior comparison CT, and the follow-up recommendation — all connected within the system.
AI can perform contextual image analysis — using the patient's clinical history, prior imaging, and current clinical question to focus analysis. AI triage prioritizes studies based on clinical urgency derived from order context. Automated follow-up tracking for incidental findings is reliable.
Implement a formal imaging ontology with structured annotations — standard vocabularies for findings (RadLex), structured radiology reports with coded measurements and follow-up recommendations, and machine-readable image annotations with spatial coordinates.
Medical images exist within a formal ontological framework. Radiology reports use structured reporting with RadLex terminology, measurements are stored as discrete values with spatial coordinates, and findings link to standard diagnostic codes. An AI agent can query 'show me all pulmonary nodules greater than 8mm found in the past year with their growth rates' and get a structured, computationally complete answer.
AI can perform sophisticated quantitative imaging analysis — measuring change over time, correlating imaging findings with pathology outcomes, and making evidence-based follow-up recommendations. Autonomous image interpretation is possible for well-defined diagnostic criteria.
Implement real-time streaming of medical images and annotations — images publish to AI consumers the moment acquisition is complete, enabling real-time AI analysis concurrent with radiologist interpretation.
Medical images stream in real-time from acquisition to AI analysis to radiologist review. The moment a CT scan completes, AI begins processing the images, generating preliminary findings, measurements, and comparison annotations before the radiologist opens the study. The imaging knowledge base is continuously updated — new studies automatically link to prior comparisons and update longitudinal tracking of measurements and findings.
Can autonomously perform preliminary image interpretation, quantitative measurement tracking, and critical finding detection in real-time. AI provides radiologists with pre-populated structured reports that they verify rather than author from scratch.
Ceiling of the CMC framework for this dimension.
Capabilities That Depend on Medical Image
Other Objects in Clinical Operations & Patient Care
Related business objects in the same function area.
Patient Record
EntityThe comprehensive longitudinal record of a patient's medical history, diagnoses, treatments, allergies, medications, and care episodes maintained by the healthcare organization.
Clinical Note
EntityThe structured or unstructured documentation of a patient encounter including SOAP notes, H&P, progress notes, and discharge summaries created by clinicians.
Vital Signs Record
EntityThe timestamped measurements of patient physiological parameters including heart rate, blood pressure, respiratory rate, temperature, and oxygen saturation.
Medication Order
EntityThe prescriber's documented instruction for a medication including drug, dose, route, frequency, duration, and clinical indication tied to a specific patient.
Laboratory Result
EntityThe structured output of clinical laboratory tests including values, reference ranges, abnormal flags, and collection timestamps for blood, urine, and other specimens.
Care Plan
EntityThe documented treatment plan for a patient including goals, interventions, responsible providers, and target outcomes for acute or chronic conditions.
Clinical Protocol
RuleThe standardized clinical pathway or evidence-based protocol defining appropriate care steps, decision points, and interventions for specific conditions or procedures.
Surgical Case Record
EntityThe comprehensive record of a surgical procedure including preoperative assessment, operative notes, anesthesia record, complications, and post-operative orders.
Clinical Workflow Template
EntityThe defined sequence of clinical tasks, handoffs, and decision points for specific care settings including ED throughput, OR turnover, and inpatient discharge.
Remote Monitoring Data Stream
EntityThe continuous or periodic data from remote patient monitoring devices including wearables, home sensors, and connected medical devices transmitted to the care team.
What Can Your Organization Deploy?
Enter your context profile or request an assessment to see which capabilities your infrastructure supports.