TABLE IV.
Data Description of the CT scans used in our experiments.
| Source | # of patients | # of scans | Slice Thickness | Median Age± SD (years) | Gender | Notes | |
|---|---|---|---|---|---|---|---|
| HPIV | NIH | 40 | 56 | 5 mm | 39.98 ±17.54 | M=18, F=22 | Low-resolution scans of patients with human para- influenza virus. Scans include cases of heavy consolidation, various other pathologies, and effusion. |
| NTB | NIH | 30 | 71 | 5 mm | 42.4±15.4 | M=17, F=13 | Patient scans with neurotic tuberculosis. Scans include cases of heavy consolidation, and cavities. |
| DAH | NIH | 26 | 68 | 5 mm | 39.23 ±18.21 | M=14, F=12 | Low-resolution scans of patients with diffuse alveolar hemorrhage. Scans include cases of heavy consolidation, and effusion. |
| TB | Public | 80 | 93 | 2.5 mm | 41.5 ±17.26 | M=49, F=31 | Patient scans with pulmonary tuberculosis. The data contains various pathologies prominently: consolidation, GGO, cavity, bleb, and pleural effusion. (http://tuberculosis.by/). |
| H1N1 | NIH | 4 | 6 | 5 mm | 50±12.8 | M=3, F=l | Patient with influenza H1N1 pneumonia. The scans show blood clot and various pathologies such as GGO, micronodules, and consolidation. |
| LOLA Challenge Data | Public | 55 | 55 | < 1.5 mm | n/a‡ | US+=55 | Data set designed as part of the lung segmentation challenge (MICAII 2011). The pathologies in lung scans range from mild to severe. (http://lola11.com). |
| EXACT Challenge Data | Public | 40 | 40 | 1 mm | 45.67 ±32.92 | M=19, F=14, US+=7 | High-resolution scans with mild pathologies (consolidation, and effusion). (http://image.diku.dk/exact). |
| Control | NIH | 14 | 14 | 2 mm | 50±17.11 | M=5, F=9 | High resolution lung scans of healthy individual. The lung scans does not show any abnormality. |
| Total | — | 288 | 403 | — | 41.91 ±20.72 | M=124, F=100, US†=61 | |
†
unspecified.
‡
data anonymized.