Cardiac mass differential on cardiac magnetic resonance. Academic tool that ranks the differential and shows the origin of each frequency. It does not diagnose.
Each item has three states: Yes, seen; Not seen, looked for but not identified (counts against the entities that usually have it); Not evaluated, does not count. For single-choice questions, choose one value or “Not evaluated”.
The catalog has 20 entities and does not cover the whole differential. It does not include, among others, vegetations, abscesses, hydatid cyst, ventricular pseudoaneurysm, coronary artery aneurysm, extrinsic compression (hiatal hernia), sarcoidosis, inflammatory myofibroblastic tumor, pericardial mesothelioma or other sarcomas such as leiomyosarcoma or synovial sarcoma. Nor does it include tuberculous pericarditis or endomyocardial fibrosis, which are common in regions where they are endemic; thrombus in Chagas cardiomyopathy or Behçet disease is ranked as thrombus. The adult base rate comes from a CMR cohort from four US centers (Shenoy 2022); the rate for adults with known cancer is derived from it and from how often each mass has an oncologic history, and the pediatric rate comes from a series of 31 centers in 6 countries. Published CMR series of cardiac masses from Latin America, Asia and Africa are small and do not yet support a base rate of their own.
Frame color. The lesion class, from anatomic variants to metastases.
Stars. Rarity, from 1 to 8, among the masses referred for CMR. More stars mean a rarer entity, which needs more evidence to reach the top of the differential.
Glow. The same rarity at a glance. Common entities do not glow; uncommon ones barely glow; rare ones glow more; very rare ones glow across the whole card.
Compartment. The circle at the top: intracavitary (CAV), intramural (MYO), pericardial (PER) or valvular (VAL).
Art and signal. A diagram of the four-chamber view with the mass at its typical site and, in one corner, a CMR thumbnail on the sequence that best shows the entity, labeled at the top as on the scanner console. The strip below is its gray-scale profile: dark, low signal or enhancement; light, high; gradient, heterogeneous; light edges around a dark center, rim enhancement. Two values separated by a slash are reported equally often; “varies” means that neither predominates.
Band. How compatible the entity is with the marked findings, relative to the top-ranked entity: compatible (at least one third as likely), possible (at least 1/20), poorly compatible (at least 1/150) and very poorly compatible. No entity leaves the differential: low-band entities go face down to the “Poorly compatible” pile, which opens to show each one with its reason; their cards are dimmed. The top-ranked entity is always compatible because the band is relative; if the marked findings fit it poorly, a warning appears.
Up and down. On the front, the findings that raise and lower the entity most against its closest rival: the top-ranked entity or, for the top-ranked entity, the second.
Show more. Below the front of the enlarged card: each finding, with the factor by which it favors the entity over its rival, its two frequencies as “x of every 100” and the origin of each figure: DATA (published figure, k of n), CLASS (comparative series of its class), V2 (provisional conversion of the weight from the previous version), GROUP (mean of its group; no data of its own), ASSUMED (frequency set by the model for findings not described in the entity), CASES (only isolated cases exist), BASE RATE (derived from the base rate of each population), and VETO (explicit contradiction: 1 of every 100 or fewer, never zero).
Stability. The order is recalculated with the model’s figures varied within their uncertainty: “stable” if the top-ranked entity holds its place, “probable” if it usually holds it but not in every recalculation, and “cannot be separated” if the figures cannot decide which comes first. This is stability under the model’s assumptions, not external validation; the same line states how much of the difference between the first two comes from published frequencies.