raw image
Summary — Dice vs label budget
| Dataset | Modality | LLM (0 train) | NN N=10 | N=25 | N=50 | N=100 | Crossover |
|---|---|---|---|---|---|---|---|
| ISIC 2018 | Dermoscopy | 0.882 | 0.749 | 0.830 | 0.828 | 0.858 | never (≤100) |
| Kvasir-SEG | Polyp / endoscopy | 0.580 | 0.628 | 0.777 | 0.808 | 0.845 | N≈10 |
| Chest X-ray | Chest X-ray | 0.788 | 0.925 | 0.943 | 0.950 | 0.956 | N≈10 |
| Spleen | Abdominal CT | 0.420 | 0.537 | 0.686 | 0.838 | 0.885 | N≈10 |
Red cells = the LLM (zero training labels) matches or beats the trained net at that budget. NN = U-Net/resnet34, ImageNet-pretrained.
ISIC 2018 (dermoscopy)
Dermoscopy
| N | LLM | NN |
|---|---|---|
| 10 | 0.882 | 0.749 |
| 25 | 0.882 | 0.830 |
| 50 | 0.882 | 0.828 |
| 100 | 0.882 | 0.858 |
The training-free LLM (0.882) beats the U-Net at every budget up to N=100 — the standout positive case (even the strongest pretrained net, SegFormer, only reaches 0.875 at N=100).
GTLLMNN @N
✓ Good cases
isic · ISIC_0036159
ground truth
LLM
0.96
NN N=10
0.91
NN N=25
0.91
NN N=50
0.95
NN N=100
0.94
isic · ISIC_0023694
raw image
ground truth
LLM
0.95
NN N=10
0.77
NN N=25
0.92
NN N=50
0.93
NN N=100
0.95
isic · ISIC_0023836
raw image
ground truth
LLM
0.95
NN N=10
0.86
NN N=25
0.96
NN N=50
0.96
NN N=100
0.97
✗ Bad cases
isic · ISIC_0023306
raw image
ground truth
LLM
0.09
NN N=10
0.54
NN N=25
0.30
NN N=50
0.15
NN N=100
0.04
isic · ISIC_0022313
raw image
ground truth
LLM
0.08
NN N=10
0.36
NN N=25
0.35
NN N=50
0.71
NN N=100
0.52
isic · ISIC_0022007
raw image
ground truth
LLM
0.06
NN N=10
0.04
NN N=25
0.13
NN N=50
0.10
NN N=100
0.15
Kvasir-SEG (polyp / endoscopy)
Polyp / endoscopy
| N | LLM | NN |
|---|---|---|
| 10 | 0.580 | 0.628 |
| 25 | 0.580 | 0.777 |
| 50 | 0.580 | 0.808 |
| 100 | 0.580 | 0.845 |
The NN leads from N=10 and runs away — the lobed, gradual-boundary polyp caps the LLM near 0.58 (≈45% of draws mislocate, ~0.35 shape residual).
GTLLMNN @N
✓ Good cases
kvasir · cju7fq7mm2pw508176uk5ugtx
raw image
ground truth
LLM
0.90
NN N=10
0.79
NN N=25
0.82
NN N=50
0.94
NN N=100
0.92
kvasir · cju5o1vu9gz8a0818eyy92bns
raw image
ground truth
LLM
0.87
NN N=10
0.72
NN N=25
0.95
NN N=50
0.93
NN N=100
0.96
kvasir · cju7amjna1ly40871ugiokehb
raw image
ground truth
LLM
0.85
NN N=10
0.69
NN N=25
0.91
NN N=50
0.96
NN N=100
0.96
✗ Bad cases
kvasir · cju3xl264ingx0850rcf0rshj
raw image
ground truth
LLM
0.12
NN N=10
0.72
NN N=25
0.70
NN N=50
0.96
NN N=100
0.95
kvasir · cju5u8gz4kj5b07552e2wpkwp
raw image
ground truth
LLM
0.10
NN N=10
0.13
NN N=25
0.56
NN N=50
0.01
NN N=100
0.42
kvasir · cju884985nlmx0817vzpax3y4
raw image
ground truth
LLM
0.06
NN N=10
0.26
NN N=25
0.55
NN N=50
0.88
NN N=100
0.85
🧪 Skill: gated polyp ROI contrast-enhance new
Bad Kvasir cases mislocate — the draw lands on the wrong region. A GT-free contrast enhancer (specular-inpaint + CLAHE-L + saturation) makes the polyp salient enough to find. It rescues mislocations but distorts already-good boundaries, so it is gated on K-draw self-consistency (applied only to low-confidence draws). Skill
polyp_roi_enhance (b45cd77).| case group | LLM raw | LLM enhanced | Δ |
|---|---|---|---|
| bad | 0.094 | 0.434 | +0.340 |
| good | 0.874 | 0.729 | -0.145 |
| gated (raw on good, enhance on bad) | 0.484 | 0.654 | +0.170 |
GTLLM rawLLM on enhanced view
✗ Bad cases — rescued by enhancement
kvasir · cju3xl264ingx0850rcf0rshj
raw
enhanced view
GT
LLM raw
0.12
LLM enhanced
0.63
kvasir · cju5u8gz4kj5b07552e2wpkwp
raw
enhanced view
GT
LLM raw
0.10
LLM enhanced
0.61
kvasir · cju884985nlmx0817vzpax3y4
raw
enhanced view
GT
LLM raw
0.06
LLM enhanced
0.06
✓ Good cases — enhancement not needed (gated off)
kvasir · cju7fq7mm2pw508176uk5ugtx
raw
enhanced view
GT
LLM raw
0.90
LLM enhanced
0.65
kvasir · cju5o1vu9gz8a0818eyy92bns
raw
enhanced view
GT
LLM raw
0.87
LLM enhanced
0.81
kvasir · cju7amjna1ly40871ugiokehb
raw
enhanced view
GT
LLM raw
0.85
LLM enhanced
0.73
Chest X-ray (lung fields)
Chest X-ray
| N | LLM | NN |
|---|---|---|
| 10 | 0.788 | 0.925 |
| 25 | 0.788 | 0.943 |
| 50 | 0.788 | 0.950 |
| 100 | 0.788 | 0.956 |
Lung fields are consistent anatomy — a NN nails them from just 10 labels (0.93). The LLM also segments them well (0.79) but never leads: the target is too learnable.
GTLLMNN @N
✓ Good cases
lungseg · MCUCXR_0150_1
raw image
ground truth
LLM
0.86
NN N=10
0.89
NN N=25
0.91
NN N=50
0.91
NN N=100
0.93
lungseg · CHNCXR_0330_1
raw image
ground truth
LLM
0.85
NN N=10
0.96
NN N=25
0.97
NN N=50
0.98
NN N=100
0.98
lungseg · CHNCXR_0421_1
raw image
ground truth
LLM
0.85
NN N=10
0.94
NN N=25
0.95
NN N=50
0.96
NN N=100
0.95
✗ Bad cases
lungseg · CHNCXR_0229_0
raw image
ground truth
LLM
0.71
NN N=10
0.91
NN N=25
0.94
NN N=50
0.94
NN N=100
0.94
lungseg · MCUCXR_0055_0
raw image
ground truth
LLM
0.71
NN N=10
0.96
NN N=25
0.97
NN N=50
0.97
NN N=100
0.98
lungseg · CHNCXR_0027_0
raw image
ground truth
LLM
0.70
NN N=10
0.94
NN N=25
0.95
NN N=50
0.95
NN N=100
0.96
🧪 Skill: X-ray CLAHE + non-convex lung trace new
Bad lung draws came out as rounded ovals — the polygon rasterizer already supports non-convex (PIL even-odd fill), so this was a drawing issue, not a skill limit. Fix = the literature-standard CXR CLAHE enhancer (
xray_clahe) + a guide to trace the true shape (concave mediastinal border + costophrenic angle). Net-positive and non-destructive on good cases (lungs are large/consistent) → applied ungated. Part of the new modality→preprocess registry (8b535f1).| case group | LLM raw | LLM enhanced | Δ |
|---|---|---|---|
| bad | 0.707 | 0.762 | +0.055 |
| good | 0.855 | 0.846 | -0.009 |
| all (ungated) | 0.781 | 0.804 | +0.023 |
GTLLM rawLLM on enhanced view
✗ Bad cases — round ovals fixed
lungseg · CHNCXR_0229_0
raw
enhanced view
GT
LLM raw
0.71
LLM enhanced
0.78
lungseg · MCUCXR_0055_0
raw
enhanced view
GT
LLM raw
0.71
LLM enhanced
0.74
lungseg · CHNCXR_0027_0
raw
enhanced view
GT
LLM raw
0.70
LLM enhanced
0.77
✓ Good cases — unaffected
lungseg · MCUCXR_0150_1
raw
enhanced view
GT
LLM raw
0.86
LLM enhanced
0.85
lungseg · CHNCXR_0330_1
raw
enhanced view
GT
LLM raw
0.85
LLM enhanced
0.83
lungseg · CHNCXR_0421_1
raw
enhanced view
GT
LLM raw
0.85
LLM enhanced
0.86
Spleen (abdominal CT)
Abdominal CT
| N | LLM | NN |
|---|---|---|
| 10 | 0.420 | 0.537 |
| 25 | 0.420 | 0.686 |
| 50 | 0.420 | 0.838 |
| 100 | 0.420 | 0.885 |
Raw LLM draws score 0.22; the diagnose-repair soft-tissue window lifts the routed pipeline to 0.42, but the NN already leads at N=10 (0.54). Spleen case masks below are the raw draws (illustrating the SIZE-law failures).
GTLLMNN @N
✓ Good cases
spleen · spleen_8_z024
raw image
ground truth
LLM
0.76
NN N=10
0.68
NN N=25
0.49
NN N=50
0.91
NN N=100
0.96
spleen · spleen_33_z057
raw image
ground truth
LLM
0.75
NN N=10
0.72
NN N=25
0.70
NN N=50
0.81
NN N=100
0.96
spleen · spleen_6_z098
raw image
ground truth
LLM
0.68
NN N=10
0.71
NN N=25
0.90
NN N=50
0.86
NN N=100
0.96
✗ Bad cases
spleen · spleen_21_z058
raw image
ground truth
LLM
0.07
NN N=10
0.55
NN N=25
0.94
NN N=50
0.85
NN N=100
0.87
spleen · spleen_21_z041
raw image
ground truth
LLM
0.01
NN N=10
0.36
NN N=25
0.52
NN N=50
0.65
NN N=100
0.59
spleen · spleen_40_z068
raw image
ground truth
LLM
0.00
NN N=10
0.70
NN N=25
0.75
NN N=50
0.95
NN N=100
0.96