Semantic Relevance Flagging: Top 100 Papers & Authors

Identifying Papers/Authors Potentially Outside Core EM Connectomics

Date: 2026-03-31
Purpose: Flag papers and authors in top 100 rankings that may be semantically unrelated to electron microscopy connectomics

1. UNRELATED PAPERS IN TOP 100

Papers to Flag (14 total)

STRONG FLAGS — Likely Unrelated to EM Connectomics:

Rank	Title	Year	Issue	Relevance
5	Autophagy-Independent Functions of the Autophagy Machinery	2019	Cell biology, not connectomics	Remove
8	Molecular mechanisms of cell death: recommendations of the Nomenclature Committee	2018	Cell biology nomenclature, not connectomics	Remove
12	Fiji: an open-source platform for biological-image analysis	2012	General image analysis tool (not connectomics-specific)	Consider removing
29	Building connectomes using diffusion MRI: why, how and but	2017	Functional connectomics (dMRI), not structural EM	Keep (related but different modality)
30	Brain Networks in Schizophrenia	2014	Clinical psychiatry focus, not methodology	Borderline
39	Graph-based network analysis of resting-state functional MRI	2010	fMRI network analysis, not EM	Remove
41	Deep Learning	2015	General ML survey, tangential at best	Remove
45	ImageNet classification with deep convolutional neural networks	2017	Computer vision, not connectomics	Remove
46	Understanding the Emergence of Neuropsychiatric Disorders With Network…	2018	Psychiatric focus via functional MRI	Borderline
47	Magnetic Resonance Imaging and Graph Theoretical Analysis of Complex Brain Networks	2011	MRI-based networks, not EM	Remove
70	Assessment of system dysfunction in the brain through MRI-based connectomics	2013	Clinical MRI focus, not EM	Remove
86	The Anatomical Distance of Functional Connections Predicts Brain Networks in Schizophrenia	2012	Clinical psychiatry + fMRI	Remove
89	Connectomic Intermediate Phenotypes for Psychiatric Disorders	2012	Psychiatric focus, uses term “connectome” but not EM	Remove
100	Disrupted Modularity and Local Connectivity of Brain Functional Networks	2010	Functional connectivity, not structural EM	Remove

Recommendation

Remove from EM connectomics bibliography (likely ranked high due to “connectome” terminology):

Ranks: 5, 8, 12, 39, 41, 45, 47, 70, 86, 89, 100
Total: 11 papers to filter

Keep but note as cross-domain (network analysis methods):

Ranks: 29, 30, 46 (2–3 papers, borderline)

2. UNRELATED AUTHORS IN TOP 100

Authors Primarily Focused on fMRI/Functional Connectomics (NOT EM Connectomics)

Rank	Author	Papers	EM Focus	fMRI Focus	Assessment
8	Yong He	57	16 (28%)	3 (5%)	BORDERLINE — Network neuroscience, not EM
14	Martijn P. van den Heuvel	50	18 (36%)	6 (12%)	BORDERLINE — “Human Connectome” but fMRI-based
16	Stephen M. Smith	47	21 (45%)	16 (34%)	STRONG FLAG — Human Connectome Project (fMRI/dMRI lead)
20	Andrew Zalesky	42	13 (31%)	6 (14%)	BORDERLINE — Network psychiatry, not EM

Assessment

These authors work on “connectomics” but in a DIFFERENT sense:

They use fMRI + diffusion MRI (non-invasive human imaging)
They analyze functional/structural connectivity in living brains
They focus on clinical psychiatry applications
They are NOT electron microscopy connectomics researchers

Should they be included in EM connectomics bibliography?

Argument for KEEP:

Network analysis methods they develop are applicable to EM connectomics
Graph theory and centrality metrics are field-agnostic
BrainNet Viewer and other tools used in connectomics community
Work on connectome at coarse resolution (whole-brain networks)

Argument for REMOVE:

Not working on EM data or structural connectomes
Different scientific questions (functional vs. structural)
Using term “connectome” but not in EM sense
Rankings inflated by citation of “connectome” papers without EM content

Recommendation

Keep but flag with asterisk noting “functional connectomics focus”:

Ranks: 8, 14, 16, 20
Note in metadata: “fMRI/functional connectivity, not EM connectomics”
Consider: Separate “network methods” category from “EM connectomics”

3. CORE EM CONNECTOMICS AUTHORS (Top 100)

For comparison, here are researchers who ARE clearly EM connectomics focused:

Rank	Author	Papers	Research Focus
1	Jeff W. Lichtman	92	EM connectomics (Drosophila, Larva) ✓
2	Olaf Sporns	76	Network neuroscience + connectomics ✓
3	Albert Cardona	72	Drosophila EM connectomics ✓
4	Edward T. Bullmore	70	fMRI + network analysis (BORDERLINE)
5	H. Sebastian Seung	64	EM connectomics, AI for segmentation ✓
7	Gregory S.X.E. Jefferis	60	Drosophila EM connectomics ✓
9	Davi D. Bock	56	EM connectomics methods ✓
10	David C. Van Essen	53	Human brain mapping (mostly fMRI) (BORDERLINE)
11	Philipp Schlegel	52	Drosophila EM connectomics ✓
12	Edward S. Boyden	51	EM techniques development ✓
13	Gerald M. Rubin	51	Drosophila neurobiology ✓
15	Hanspeter Pfister	48	EM image analysis + AI ✓
17	Sven Dorkenwald	45	EM connectomics (Drosophila, Zebrafish) ✓
18	Moritz Helmstaedter	43	EM connectomics + theory ✓
19	Aljoscha Nern	42	Drosophila connectomics ✓

4. SUMMARY & FILTERING RECOMMENDATIONS

Papers to Filter Out (Remove from EM connectomics rankings)

High-confidence removals (clear non-connectomics):

Rank 5: Autophagy (cell biology)
Rank 8: Cell death mechanisms
Rank 41: Deep Learning (general ML)
Rank 45: ImageNet/CNNs (computer vision)

Medium-confidence removals (connectomics by fMRI, not EM):

Rank 29: dMRI connectomics (different modality, but related)
Rank 39: fMRI network analysis
Rank 47: MRI-based networks
Rank 70: Clinical MRI connectomics
Rank 86: Psychiatric MRI
Rank 89: Connectomic phenotypes
Rank 100: Functional networks

Low-confidence removals (tools/general connectomics):

Rank 12: Fiji (general image analysis tool)
Rank 30, 46: Clinical/psychiatric focus via connectomics

Authors to Flag (Not Remove, But Flag)

Functional connectomics researchers (keep but flag):

Rank 8: Yong He
Rank 14: Martijn P. van den Heuvel
Rank 16: Stephen M. Smith
Rank 20: Andrew Zalesky

Action: Add “connectomics_modality” field to author_rankings.json:

{
  "name": "Stephen M. Smith",
  "rank": 16,
  "connectomics_modality": "fMRI/dMRI (functional)",
  "note": "Human Connectome Project — not EM connectomics"
}

5. REVISED FILTERING STRATEGY

Option 1: Conservative (Remove 11 papers)

Remove papers clearly off-topic (cell biology, ML, fMRI)
Keep network methods papers
Updated ranking: top 90 papers for EM connectomics

Option 2: Moderate (Remove 4 papers)

Remove only obvious off-topic (autophagy, cell death, general ML, vision)
Keep all connectomics-related papers (including fMRI)
Updated ranking: top 96 papers

Option 3: No filtering

Keep all papers, but add “connectomics_type” metadata
Include in outputs but note modality (EM, fMRI, dMRI)
Most transparent approach

Recommended: Option 3 (Metadata Flagging)

Why:

Preserves full ranking while being transparent
Shows how “connectome” term is used across neuroscience
Users can filter themselves
Easier to justify to critics

Implementation:

Add “connectomics_type” field to papers:
- “em” = EM microscopy connectomics
- “functional” = fMRI/functional connectivity
- “dmri” = diffusion MRI connectomics
- “network_methods” = methods applicable to connectomics
- “unrelated” = cell biology, general ML, etc.
Add “connectomics_modality” field to authors:
- “em” = EM connectomics researcher
- “functional” = fMRI/network neuroscience
- “mixed” = both EM and functional
Filter at visualization layer:
- Default: show all (including borderline)
- Toggle: “EM connectomics only” (filters out functional)
- Toggle: “Core EM researcher only” (filters authors)

6. ACTION ITEMS

Add “connectomics_type” to paper_rankings_all.json
Add “connectomics_modality” to author_rankings.json
Create filtered versions:
- paper_rankings_em_only.json (EM connectomics papers)
- author_rankings_em_only.json (EM connectomics researchers)
Update visualizations to support filtering
Document filtering rationale in BIBLIOGRAPHY_ANALYSIS_DOCS.md
Prepare for methodology review discussion

Note: This analysis is data-driven (keyword + title analysis). Manual expert review recommended before finalizing filters.