O3‐09‐04: Network‐level analysis of PET metabolic features for detection of Alzheimer's disease
Bibliographic record
Abstract
FDG-PET scans offer a powerful window into the metabolic activity in the brain, enabling the characterization of disrupted metabolic patterns in Alzheimer's disease (AD). Regional decrease in metabolic activity has been shown to be a sensitive signature of dementias [1]. Network-level imaging biomarkers derived from structural MRI exhibit potential for individual diganosis of AD [2,3]. Although regional analysis of metabolic features has been studied extensively, the inter-regional metabolic co-variation has not yet been analyzed and used in designing classifiers for incipient AD. We present here a novel study into the group differences in network properties, as well as demonstrate their predictive utility in the detection of prodromal AD. FDG-PET scans were obtained from ADNI-1 dataset for 55 normal controls (NC) and 89 AD subjects at baseline and for 69 progressive MCI (pMCI) subjects (1 year prior to conversion to AD). Raw PET measures were first corrected for partial volume effect (PVE) and then normalized by the mean uptake value of brainstem [1]. They were co-registered to the corresponding T1-MRI scans, and mean uptake values were summarized in patch-wise averages [2]. Based on patch-wise mean uptake values, a GRADIENT network [3] was constructed when the inter-patch difference is greater than a predefined threshold. Network properties (nodal degree, betweenness-centrality etc) were computed from this graph (denoted as PetNet features) to represent the individual subject. They were fused via multiple kernel learning to build a predictive model. The group differences in PetNet features (Figures 1 and 2) reveal an interesting pattern between NC and AD on a group-level, with changes seen in medial temporal lobar structures and few anterior locations. Using RHsT cross-validation [2], we evaluate the classification performance of PetNet features (Figure 3). This approach resulted in an area under ROC (AUC) of 0.92 in discriminating AD from NC, and AUC of 0.85 in discriminating pMCI from NC, which compare favorably to those from structural GRADIENT features [3]. Group Differences between NC and AD in the nodal degree property of PetNet networks. This reveals lowered nodal-degree in AD as compared to NC, with changes in medial temporal lobar structures as well as some anterior brain locations. Group mean Differences between NC and AD in the betweenness centrality property of PetNet networks. These reveal increased betweenness-centrality pattern in AD in the medial temporal lobar structures. Classification results from MKL classifier based on PetNet features. We present novel network-level features summarizing the covariation of metabolism in the brain derived from FDG-PET images. We demonstrate the diagnostic utility of PetNet features for the early detection of Alzheimer's disease.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.005 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".