Table 1.
Summary of HSI applications in gastroenterology.
| Application/Disease | Spectral Range (nm) | HSI Technology | Experiment Type | Study Subjects | Data Analysis Methods (Category */Method ¥) | Reference(s) |
|---|---|---|---|---|---|---|
| Biliary tree visualization | 650–1050 | LCTF | In-vivo | Swine | D, E/PCA | [55] |
| Colon cancer detection | 400–700 | LCTF | Ex-vivo | Humans | F, E/LPM | [59,60] |
| Organs identification during surgery | 900–1700 | Push-broom | In-vivo | Swine | SA, P/DWT; C/SOM | [61] |
| Identifying tissues during surgery | 350–1830 | DRS | In-vivo | Humans | SA, F/SGAD; C/SVM | [62] |
| Tissue identification during colorectal surgery | 440–1830 | DRS | Ex-vivo | Humans | SA, C/TPCR | [63] |
| Malignant colorectal tumors and adenomatous polyps | 405–665 | Filter Wheel | In-vivo | Humans | R/RDFS; C/SVM | [64] |
| Colon cancer detection | 300–1800 | Spectroscopy | Ex-vivo | Humans | C/LDA; C/SVM | [66] |
| Oxygenation measurement (small bowel) | 400–720 | LCTF | In-vivo | Swine | Ex/Linear light model | [67] |
| Oxygenation measurement (small bowel) | 470–700 | Filter-based | In-vivo | Swine | Ex/Non-linear light model | [68] |
| Suture recommendation (intestinal anastomosis) | 470–770 | LED-based | Ex-vivo | Swine | Ex/2D-filtering, SAM and composite images from the multispectral image | [69] |
| Monitoring radiofrequency fusions in small bowel | 460–700 | LCTF | In-vivo | Swine | Ex/Linear light model | [70] |
| Biliary trees identification | 650–1100 | LCTF | In-vivo | Swine | D, E/PCA | [71] |
| Biliary anatomy visualization | 650–700 | LCTF | Ex-vivo | Swine | S/LMM, R/PCA | [72] |
| Intestinal ischemia identification | 400–1700 | Push-broom | In-vivo | Swine | I/Ischemia Index; C/SVM | [74] |
| Gastric cancer detection | 1000–2500 | Push-broom | Ex-vivo | Humans | I/Cancer Index; C/SVM | [75] |
| Gastric ulcers | 405–665 | Filter Wheel | In-vivo | Humans | R, E/DI | [76] |
| Gastric cancer | 400–800 | N/A | Ex-vivo | Humans | C/MDC | [77,89] |
| Gastric cancer | 400–650 | Tunable Light Source | In-vivo | Humans | C/SVM; C/RF; C/RobustBoost; C/AdaBoost | [78] |
| Colon cancer detection | 450–850 | Tunable Light Source | In-vitro | Humans | R/ICA; R/PCA; C/k-Means; C/LDA; C/SVM | [79,80] |
| Colon cancer detection | 440–700 | Tunable Light Source | In-vitro | Humans | F/CLBP; R/PCA; C/LDA; C/SVM | [81] |
| Gastric cancer cell identification | 420–720 | LCTF | In-vitro | Humans | R/Manual band selection; C/ANNs | [82] |
| Colonic adenocarcinoma identification | 390–700 | LCTF | Ex-vivo | Humans | SA | [83] |
| Colon cancer detection | 360–550 | LCTF | In-vitro | Humans | S/LMM; R/PCA | [84,85] |
| Colorectal cell differentiation | 400–1700 | LCTF | In-vitro | Humans | F/LBP, C/RF | [86] |
| Colon cancer detection | 400–1000 | Push-broom | Ex-vivo | Humans | DR/SPA; C/LDA | [90] |
* Categories of data analysis methods: (P) Preprocessing; (F) Feature extraction; (C) Classification; (R) Data Reduction; (S) Spectral Unmixing; (I) Normalized Difference Index; (E) Tissue Visualization Enhancement; (SA) Spectral Signature Analysis; (Ex) Exploratory Data Analysis. ¥ Data analysis methods: (SGAD) Spectral Gradients and Amplitude Differences; (SVM) Support Vector Machines; (DWT) Discrete Wavelet Transformation; (SOM) Self-Organizing Maps; (CLBP) Circular Local Binary Patterns; (PCA) Principal Component Analysis; (LDA) Linear Discriminant Analysis; (LPM) Light Propagation Modeling; (LMM) Linear Mixture Model; (ICA) Independent Component Analysis; (RDFS) Recursive Divergence Feature Selection; (DI) Dependence of Information; (MDC) Minimum Distance Classifiers; (TPCR) Total Principal Component Regression; (RF) Random Forest; (SAM) Spectral Angle Mapper; (SPA) Successive Projection Algorithm; (LBP) Local Binary Pattern; (ANNs) Artificial Neural Networks.