Resting-state studies frequently catalogue significant connections one region at a time, although diagnosis is expressed by the arrangement of abnormalities across several circuits. We asked whether hippocampal-output disturbances form a diagnostically meaningful geometry after the amount and reliability of evidence attached to each pathway are considered. A signed evidence matrix was assembled for seven hippocampal connections evaluated in 475 individuals: 117 with schizophrenia, 103 with bipolar disorder, 96 with major depressive disorder, and 159 healthy participants. Each connection was weighted by the product of its omnibus F statistic and the logarithm of its corrected cluster extent. Weighted signed distances, alteration burden, diagnostic entropy, and seed-level discrimination were calculated. An exact placement test enumerated 6,174 circuit arrangements while preserving each diagnosis’s number and direction of alterations. One-pathway deletion and alternative weighting analyses assessed stability. Schizophrenia and bipolar disorder occupied the nearest positions (distance = 0.329), whereas bipolar disorder and major depressive disorder were most distant (0.433). Their signed concordance was likewise greatest for schizophrenia and bipolar disorder (0.568). The subiculum carried 54.5% of seed-level discriminatory information, compared with 30.2% for cornuammonis and 15.4% for dentate gyrus. Total separation was 1.135, below the exact-placement mean of 1.282±0.112, but the departure was not statistically decisive (two-sided p = 0.2309). Deleting one pathway changed total separation only from 1.105 to 1.217 (coefficient of variation 3.4%). Hippocampal-output abnormalities encode a stable, compact diagnostic geometry rather than a sharply partitioned set of disorders. The principal organizational feature is a subicular distribution spanning prefrontal, temporal, and motor targets, while the short schizophrenia–bipolar distance arises from concordant anterior cingulate and orbitofrontal reductions. The method distinguishes reliable organization from apparent uniqueness without treating thresholded clusters as independent biomarkers.
Resting-state studies frequently catalogue significant connections one region at a time, although diagnosis is expressed by the arrangement of abnormalities across several circuits. We asked whether hippocampal-output disturbances form a diagnostically meaningful geometry after the amount and reliability of evidence attached to each pathway are considered. A signed evidence matrix was assembled for seven hippocampal connections evaluated in 475 individuals: 117 with schizophrenia, 103 with bipolar disorder, 96 with major depressive disorder, and 159 healthy participants. Each connection was weighted by the product of its omnibus F statistic and the logarithm of its corrected cluster extent. Weighted signed distances, alteration burden, diagnostic entropy, and seed-level discrimination were calculated. An exact placement test enumerated 6,174 circuit arrangements while preserving each diagnosis’s number and direction of alterations. One-pathway deletion and alternative weighting analyses assessed stability. Schizophrenia and bipolar disorder occupied the nearest positions (distance = 0.329), whereas bipolar disorder and major depressive disorder were most distant (0.433). Their signed concordance was likewise greatest for schizophrenia and bipolar disorder (0.568). The subiculum carried 54.5% of seed-level discriminatory information, compared with 30.2% for cornuammonis and 15.4% for dentate gyrus. Total separation was 1.135, below the exact-placement mean of 1.282±0.112, but the departure was not statistically decisive (two-sided p = 0.2309). Deleting one pathway changed total separation only from 1.105 to 1.217 (coefficient of variation 3.4%). Hippocampal-output abnormalities encode a stable, compact diagnostic geometry rather than a sharply partitioned set of disorders. The principal organizational feature is a subicular distribution spanning prefrontal, temporal, and motor targets, while the short schizophrenia–bipolar distance arises from concordant anterior cingulate and orbitofrontal reductions. The method distinguishes reliable organization from apparent uniqueness without treating thresholded clusters as independent biomarkers.
