7.2 EQUIP: QA Oversight, Clinical Image Quality, and Corrective Action

7.2 EQUIP: QA Oversight, Clinical Image Quality, and Corrective Action

Quick Answer: EQUIP (Enhancing Quality Using the Inspection Program) is how FDA inspects clinical image quality that 21 CFR 900.12 already required. Inspectors ask three questions: Does the facility correct poor clinical images? Does it review a sample of each active technologist's and each active interpreting physician's images against its accreditation body's standards, with documentation since the last inspection? Does the lead interpreting physician (LIP) oversee QA/QC records and corrective actions? FDA does not publish a "every Tuesday" review calendar. Facilities design a process; inspectors look for a real system and, for the sample review, dated documentation at least once per inspection cycle.

ARRT added EQUIP to Image Production (mammographic technique and image evaluation) in the specifications implemented September 1, 2025. The exam is not asking you to recite a vendor in-service. It is asking who owns image quality, what happens when an image fails, and which attributes the accreditation body and EQUIP both use.

Why EQUIP exists

FDA has long had regulations that clinical images must continue to meet the facility's accreditation-body standards (21 CFR 900.12(i)), that interpreting physicians follow facility procedures when images are of poor quality (21 CFR 900.12(d)(1)(ii)(A)), and that the LIP ensure QA/QC records — monitoring data, problems, corrective actions, and effectiveness — are maintained (21 CFR 900.12(d)(2)). Before EQUIP, annual inspections did not systematically ask about those clinical-image rules. EQUIP put three question sets into the inspection so that continuous clinical image quality, not only phantom scores, is visible.

FDA does not hand facilities a sample program. Each site builds a process that fits its volume and staffing. The LIP is the MQSA-designated person most responsible for ensuring image quality is maintained. FDA is explicit that "Lead Interpreting Physician" is an MQSA designation and does not confer supervisory authority over other interpreting physicians. The LIP is still an IP, so a sample of images the LIP accepted for interpretation must be in the facility review.

Pillar 1 — Quality assurance / clinical image corrective action

Inspection question 1: Does the facility have procedures for corrective action when clinical images are of poor quality?

Sub-questions FDA actually uses:

• Is there a mechanism for ongoing interpreting-physician feedback on image quality to radiologic technologists or other designated personnel?
• Do procedures require documenting corrective actions and documenting the effectiveness of those actions?

The interpreting physician decides, for this inspection question, whether images are of poor quality. Software tools may help, but the IP owns the judgment. The facility chooses the feedback mechanism: a rejected-image log, a workstation flag, a huddle note, a secure message. FDA will not grade the stationery. FDA will check that a system exists with those two elements.

A written SOP is not required for this pillar. Staff may explain the system verbally. If no images were poor, FDA does not require a document that says "no corrective action." FDA also does not set a numeric cap on how many corrective-action cases are "too many"; the facility judges effectiveness. There is no FDA-mandated clock such as "repeat the patient before lunch" and no required retention period for IP-to-RT image-quality feedback. The clinical goal FDA names is a high-quality individual mammogram that can be interpreted.

What failure looks like on a real shift. An IP opens an MLO and the inframammary fold is not on the image, the pectoral muscle is short, and the posterior nipple line is short compared with the CC. That is a positioning failure under the AB attributes. Corrective action is not "send it anyway and mention it in the report." The IP feeds back to the RT. The RT's action might be repeating the view with a different patient stance, a different paddle, or additional time for the arm and abdomen. Effectiveness is the next image (or a documented sample of later work) showing the missing anatomy. Compression that is inadequate for the breast thickness, exposure that saturates or under-drives the detector, motion blur, and a missing laterality marker are the same loop: IP flags, RT corrects, someone records that the correction worked.

Pillar 2 — Clinical image quality (the sample review)

Inspection question 2: Does the facility have procedures to ensure clinical images continue to comply with the AB's clinical image quality standards?

Sub-questions:

• Is there a mechanism for regular reviews of image-quality attributes of a sample of mammograms performed by each active RT and a sample accepted for interpretation by each active IP?
• Is there documentation of such a review since the last inspection?

This is the only EQUIP pillar for which FDA requires written documentation. A verbal claim is not enough. Acceptable documentation can be a summary report, an LIP statement that a review was performed, meeting records, or memos of results to RTs and IPs. The review must be dated. It need not be signed. Sample size is the facility's choice. Every active RT and every active IP at that certified facility must be in the sample, including diagnostic-only staff and including the LIP. MQSA does not split screening from diagnostic for this review. Personnel who have already left need not be included. A technologist may not perform the periodic review on her own images. An IP (including a sole IP who is also the LIP) may review his or her own accepted images, though FDA notes that another IP is useful when one is available.

Frequency FDA will own in writing: because inspectors ask for documentation since the last annual inspection, the periodic review must occur at least annually. FDA encourages more frequent review (monthly or quarterly). FDA does not require a named weekday. Daily interpretation of every mammogram does not count as the periodic sample review. Repeat/reject analysis does not count. Images older than a year, from before the previous inspection, should not be pulled. Multi-site networks must review each certified facility separately, though they may put clearly labeled facilities on one report. Peer review may count only if it also assesses whether images accepted for interpretation meet AB image-quality standards. The review is not a test of interpretive accuracy; that belongs to the medical outcomes audit.

Any designated person, group, or organization working in conjunction with an IP may perform the review. The LIP remains ultimately responsible for 21 CFR 900.12(i).

Pillar 3 — Quality control oversight by the LIP

Inspection question 3: Does the facility have a procedure for LIP oversight of QA/QC records and corrective actions?

The procedure must include LIP oversight of QA/QC records, including review of whether all required tests were performed at the required frequency, and LIP review of whether appropriate corrective actions were performed when needed. Tests performed by the QC technologist, the medical physicist, and any other designated person are in scope.

On inspection day the facility shows this pillar in one of three ways: the LIP attests in writing, the LIP answers questions in person or by phone, or the facility presents an SOP signed by the LIP. If the LIP is off-site or in a teleradiology group, oversight is still that LIP's MQSA job. The LIP may designate someone to perform the record review; the LIP still attests. FDA does not require a fixed LIP-review calendar and does not require extra QA/QC training for the LIP. A new attestation or a newly signed SOP is needed each year the LIP is not available to speak with the inspector.

Repeat Level 2 EQUIP observations can send the facility to its AB for additional mammography review (AMR). Take EQUIP citations as image-quality problems, not paperwork theater.

The eight attributes — EQUIP and Image Production E

FDA EQUIP FAQ Q2.12 points to eight attributes in 21 CFR 900.4(c)(2)(i–viii). The same list is the clinical image-quality language ARRT tests under Image Production E (evaluation of image quality). Use it for EQUIP reviews and for deciding whether to repeat a view before the patient leaves.

FDA states these attributes apply to screen-film, FFDM, and DBT. There is no separate "3D-only" EQUIP attribute list.

Tie to ARRT clinical experience

ARRT's Mammography Clinical Experience Requirements (Board approved July 2019; implementation July 1, 2020, still the clinical document ARRT publishes for this pathway) list Mammographic Image Evaluation as one documented procedure: the candidate must consult with an MQSA-qualified interpreting physician to review at least 10 mammographic cases for breast anatomy, pathology, and image quality, and establish corrective action per EQUIP. That consult is the same conversation EQUIP expects between IP and RT, done deliberately with a qualified IP rather than guessed from a textbook atlas. Use the eight-attribute table in that review. Logging ten cases without talking through failed positioning or a missing marker does not meet the spirit of the requirement ARRT wrote.

Exam trap: EQUIP does not replace manufacturer QC or the physicist's annual survey. Phantom passing and clinical images failing can both be true. EQUIP asks whether someone with interpretive responsibility looked at patient images and whether the LIP looked at whether QC was actually done.