File size: 2,893 Bytes
e043529
 
 
e38da7e
e043529
 
 
 
 
 
 
 
 
 
 
 
 
 
68156be
e38da7e
e043529
 
 
 
ac41c0b
e043529
 
 
 
 
 
 
 
 
 
 
 
 
68156be
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ac41c0b
e043529
 
 
 
 
 
 
 
 
 
 
 
68156be
e043529
68156be
e043529
68156be
 
 
e043529
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
---
license: other
license_name: ohsu-non-commercial
license_link: https://github.com/ChangLab/miniMTI/blob/publication/LICENSE
tags:
  - biology
  - multiplex-imaging
  - computational-pathology
  - cycif
  - colorectal-cancer
size_categories:
  - 1K<n<10K
---

# miniMTI-CRC Example Data

Example single-cell imaging data for testing [miniMTI](https://huggingface.co/changlab/miniMTI-CRC), a molecularly anchored virtual staining framework for multiplex tissue imaging panel reduction.

**Paper:** [bioRxiv 2026.01.21.700911](https://www.biorxiv.org/content/10.64898/2026.01.21.700911v1)  
**Code:** [GitHub](https://github.com/ChangLab/miniMTI)  
**Model:** [changlab/miniMTI-CRC](https://huggingface.co/changlab/miniMTI-CRC)

## Dataset Description

10,000 single-cell image patches randomly sampled (seed=42) from CRC-Orion sample CRC04 (colorectal cancer tissue WSI, RareCyte Orion platform).

### File

- `example_CRC04_10k.h5` — HDF5 file (~178 MB)

### HDF5 Structure

| Dataset    | Shape              | Type   | Description |
|------------|--------------------|--------|-------------|
| `images`   | (10000, 32, 32, 20)| uint8  | 17 IF channels + 3 H&E (RGB) channels |
| `masks`    | (10000, 32, 32)    | bool   | Binary cell segmentation masks |
| `metadata` | (10000,)           | string | Cell IDs and coordinates: `<sample>-CellID-<id>-x=<x>-y=<y>` |

### Channel Ordering (20 raw channels)

| Index | Channel     |
|-------|-------------|
| 0     | DAPI        |
| 1     | CD31        |
| 2     | CD45        |
| 3     | CD68        |
| 4     | CD4         |
| 5     | FOXP3       |
| 6     | CD8a        |
| 7     | CD45RO      |
| 8     | CD20        |
| 9     | PD-L1       |
| 10    | CD3e        |
| 11    | CD163       |
| 12    | E-cadherin  |
| 13    | PD-1        |
| 14    | Ki67        |
| 15    | PanCK       |
| 16    | aSMA        |
| 17    | H&E (R)     |
| 18    | H&E (G)     |
| 19    | H&E (B)     |

Channels 0–16 are immunofluorescence markers. Channels 17–19 are co-registered H&E RGB. The miniMTI model treats each IF channel as a separate marker and the three H&E channels as a single marker (18 markers total).

## Usage

```python
from huggingface_hub import hf_hub_download

# Download example data
path = hf_hub_download(
    repo_id="changlab/miniMTI-CRC-example",
    filename="example_CRC04_10k.h5",
    repo_type="dataset",
)
```

```bash
# Run inference with miniMTI
python scripts/inference_example.py \
    --val-file $path \
    --input-channels 17,6,11,13
```

## Citation

```bibtex
@article{sims2026minimti,
  title={miniMTI: minimal multiplex tissue imaging enhances biomarker expression prediction from histology},
  author={Sims, Z. and Govindarajan, S. and Ait-Ahmad, K. and Ak, C. and Kuykendall, M. and Mills, G. B. and Eksi, E. and Chang, Y. H.},
  journal={bioRxiv},
  year={2026},
  doi={10.64898/2026.01.21.700911}
}
```