Medical Imaging Quality Improvement Framework
From Imaging Problem to Sustained Improvement
OraDigit helps your organization define an imaging-quality or operational problem, understand current performance, investigate contributing factors, design an appropriate change and monitor whether improvement lasts.
We begin with the problem. Quality improvement is the service; data, technology, products and research support the work where they are appropriate.
Revisit the problem and baseline when conditions change.
This is the detailed operating model behind Our Approach: Assess covers Discover and Baseline; Identify covers Analyze; Improve, Measure and Sustain carry through. Solutions describes the imaging improvement domains we work in, rather than six separate services.
01 · Discover
Define the problem with the people involved.
What is happening, where, how often and with what consequence? Agree the modality, site, scanner, workflow stage and process boundary. Identify the people affected, the stakeholders needed and whether the concern involves quality, safety, workflow, consistency, readiness, measurement or technology.
Separate the observed problem from a suspected cause. A pattern of repeat studies is an observation; inadequate preparation is a hypothesis to investigate. Discovery does not establish the root cause.
Possible output: an agreed improvement question, problem statement, scope, stakeholder map and preliminary workflow map or measurement hypothesis.
02 · Baseline
Establish measurable current performance.
Agree a metric definition, numerator, denominator and measurement period. Record inclusion and exclusion rules, data source and data-quality limitations. Review variation and the current trend in operational context; stratify by modality, site, scanner, protocol or process factor where useful.
A count of repeated scans alone is not enough. A repeat rate may use repeat studies / eligible studies over a defined period, with consistent eligibility and recording rules. Missing or inconsistent data may require better measurement before an intervention can be evaluated.
Relevant measures might include repeat/reject, nondiagnostic or motion/artifact rates; preparation success, including heart-rate preparation where relevant; protocol adherence or variation; cancellations, rescheduling, wait or turnaround time; throughput, utilization or capacity; QC findings, corrective-action completion or readiness findings. Not every metric fits every organization.
Possible output: a baseline metric specification, performance analysis and data-quality findings.
03 · Analyze
Investigate variation and contributing factors.
Explore preparation, scheduling, protocol design and adherence, technologist workflow, training and competency, equipment, reconstruction, communication and handoffs, ordering and documentation, staffing, data visibility, standardization, QC follow-up, interoperability and technology limitations as relevant to the question.
Contributing factors are investigated; they should not be declared causes without evidence. Process mapping, workflow observation, stratification, trend, Pareto and variation analysis, run/control charts, cause-and-effect analysis, data-quality assessment or protocol/process comparison may help. Select methods that fit the data and question.
Retrospective correlations may show association rather than causality. Record uncertainty and alternative explanations before prioritizing an intervention with the organization.
Possible output: contributing-factor analysis and prioritized improvement opportunities, with supporting evidence and limitations.
04 · Improve
Choose an intervention that follows the evidence.
Sometimes the appropriate improvement is a process change. Sometimes it is better measurement. Sometimes technology helps. Technology is not the default intervention: the intervention should follow the problem and evidence.
Options may include preparation changes, protocol/process standardization, workflow redesign, communication or scheduling changes, QC follow-up, training, documentation and performance feedback. Data visibility, dashboards, automation, interoperability, decision support, software or AI/ML may be useful when justified by the task.
Agree owners, review points, expected effects and a measurement plan before implementation. The organization approves and controls changes through its clinical and operational governance.
Possible output: an intervention plan and agreed workflow or technology requirements.
05 · Measure
Check whether performance actually improved.
Implementation alone is not evidence of improvement. Compare results with the baseline using the same metric definition, a comparable denominator and an appropriate observation period. Review trends over time and relevant site, scanner or workflow strata, with changes in context recorded.
Use process, outcome and balancing measures to look for unintended consequences. Higher throughput should not quietly increase repeat rates, safety concerns or downstream delays. Better preparation and image quality may affect appointment duration or scheduling. Technology needs evaluation for workflow burden and operational reliability as well as functionality.
Installing a dashboard or deploying software is an implementation. Improvement requires evidence that the defined process or outcome changed in the intended direction without unacceptable unintended effects. Do not infer causal attribution or statistical significance unless the study design supports it.
Possible output: a measured comparison, its limitations and a decision to continue, adapt or stop the change.
06 · Sustain
Make performance visible and respond to drift.
Agree ongoing KPI monitoring, trend review, thresholds or triggers and control limits where appropriate. Connect QC follow-up and corrective-action tracking with named owners, standard processes and protocols, documentation and periodic review.
Detect regression and reassess when performance drifts or conditions change. The aim is to see performance, know when it changes and respond appropriately. A one-time improvement is not the end of the work; monitoring arrangements and any ongoing support require agreement.
Possible output: a monitoring plan, responsibilities and sustainment recommendations.
Data & Technology Across the Improvement Cycle
Data & Technology is cross-cutting enablement, not a seventh phase. Workflow data gathering can support Discover; extraction, normalization and KPI computation can support Baseline; stratification, statistical analysis, workflow analytics and data-quality review can support Analyze.
Automation, interoperability, software engineering and decision support may enable Improve, including AI and ML where justified. Dashboards, automated KPI pipelines and comparison/trend analysis may support Measure; analytics, alerts, monitoring, QC tracking and performance visibility may support Sustain.
Data platforms and MLOps can support governed data and model operations where needed. Each capability needs a defined purpose, suitable evidence, appropriate governance and evaluation against the improvement question.
Products & Research
Products and research may support specific improvement questions where their maturity, evidence and intended use are appropriate. Order Helper and AI Assistant remain Demonstrations. PET Quant and PET Response Tracker remain Research prototypes. These classifications do not establish validated clinical effectiveness. Research & Technology describes the current work and its limits.
How an Engagement May Work
The Imaging Quality Improvement Assessment is a structured entry point for Discover and Baseline. Together, we agree scope, responsibilities, evidence needs and review points before work proceeds. The phases guide the engagement; findings may require returning to an earlier question.
OraDigit may support problem definition, workflow analysis, KPI design, measurement, analytics, improvement design, technology enablement and monitoring. The organization retains clinical and operational governance. Depending on scope, participation may include an executive or operational sponsor, imaging and quality leadership, frontline staff and IT/data support, with appropriate aggregate operational data and authority to implement agreed changes.
Depending on scope, an engagement may produce a defined problem, current-state workflow, baseline metric specification and performance analysis, data-quality findings, contributing-factor analysis, prioritized opportunities, an intervention plan, KPI/dashboard specification, workflow or technology requirements, a measurement and monitoring plan, and sustainment recommendations. Not every engagement produces every artifact; scope and outputs require agreement.
How Improvement Is Demonstrated
A metric needs a definition. A baseline needs a denominator and time period. Understand variation before declaring a problem solved, and do not automatically describe association as causation. Evaluate an intervention after implementation, monitor whether improvement lasts and include balancing measures: fixing one process can create another problem.
Accreditation & Compliance Readiness
Readiness work can help organize processes, documentation, measurement and identified gaps for the organization's own review and applicable external requirements. OraDigit does not grant accreditation, certify compliance, provide regulatory approval or guarantee accreditation outcomes.
Clinical and information boundaries
This framework concerns organizational quality improvement. It does not diagnose patients, select individual treatment, provide patient-specific medical advice, replace qualified clinical judgment or function as a medical device. Clinical and operational decisions remain with the responsible organization and qualified professionals.
This page collects no information. Assessment and Contact use their existing inquiry processes; do not submit patient-identifying information, PHI, credentials or confidential datasets.
Start with the imaging problem you need to understand.
Define the concern, current performance and improvement question. The assessment helps organize that starting point.