In recent years, access to health information has become practically unlimited. Upon receiving a test result, many patients immediately turn to the internet in search of answers. However, when the subject involves systems such as Bi-rads, Ti-rads or Pi-rads, this practice requires caution.
These classifications are part of imaging exams and are used to describe findings and estimate the risk of malignancy based on technical criteria observed in ultrasounds and magnetic resonance imaging. They indicate the probability of a change being benign or requiring additional investigation and guide clinical management. When interpreted outside of this context, they can generate mistaken readings, unnecessary anxiety and even harmful decisions, such as postponing an appointment, starting supplements on your own or interrupting a treatment that is already in progress.
These systems were developed to standardize communication between healthcare professionals and qualify clinical decision-making. A report organized by these classifications allows different doctors to understand the same finding based on objective criteria and share a common basis for evaluation. Even so, the result of the exam is not sustainable in isolation and needs to be analyzed in conjunction with clinical data that do not appear in the report, such as the patient’s history, symptoms, age, risk factors and previous exams.
Limits of interpretation outside the clinical context
The search for quick answers outside the clinical environment has been the subject of recent studies. Research carried out by the University of Oxford, published by the scientific journal Nature Medicine in February 2026, identified important limitations in the use of artificial intelligence tools for interpreting medical information. The results point to significant flaws in the accuracy of responses and difficulties in offering adequate guidance to the lay public, which reinforces the need for .
In the field of radiology, Bi-rads are used to evaluate breasts in exams such as mammography, ultrasound and magnetic resonance imaging. Ti-rads is applied to the analysis of thyroid nodules by ultrasound, considering characteristics such as composition and contours. Pi-rads guides the interpretation of , especially in contexts of clinical suspicion. These systems were structured to estimate risk and suggest actions, not to establish definitive diagnoses.
The categories that make up these systems range from 1 to 5 and indicate progressive levels of suspicion based on defined technical criteria:
- Results classified as 1 and 2 correspond to benign findings, which generally do not require immediate intervention, but can be monitored according to medical advice;
- Category 3 indicates a low probability of malignancy and usually requires closer monitoring, with reevaluation at a defined interval;
- Categories 4 and 5 indicate a greater degree of suspicion and, in most cases, lead to the indication of additional investigation to clarify the diagnosis.
Why Isolated Interpretation Can Lead to Errors
It is important to consider that these classifications do not in themselves establish a definitive diagnosis. Confirming or excluding the disease depends on additional evaluation steps, which may include additional tests and, when indicated, procedures such as biopsy. The isolated interpretation of the category can lead to inappropriate conclusions about the severity of the condition.
Given any result, especially in the highest categories, the safest course of action involves evaluation by a specialist. Proper reading of the exam requires integration between the report and the patient’s clinical context, including health history, age, symptoms, previous exams and risk factors. This analysis allows for more accurate decisions about the need for follow-up, repeat examination or additional investigation.
Imaging exams are part of a broader diagnostic process and should not be understood as a conclusion in itself. Technological advances have expanded detection and analysis capabilities, contributing to greater precision in medical practice. Still, the use of these tools requires qualified interpretation. Technological resources can support access to information, but they do not replace the training, experience and responsibility involved in conducting health care.
*Text written by doctor Giovanni Guido Cerri, president of the Councils of the Institutes of Radiology (INRAD) and Technological Innovation (InovaHC), of the Hospital das Clínicas of FMUSP
