Evaluation of Preprocessing Methods on Independent Medical Hyperspectral Databases to Improve Analysis

Abstract

Currently, one of the most common causes of death worldwide is cancer. The development of innovative methods to support the early and accurate detection of cancers is required to increase the recovery rate of patients. Several studies have shown that medical Hyperspectral Imaging (HSI) combined with artificial intelligence algorithms is a powerful tool for cancer detection. Various preprocessing methods are commonly applied to hyperspectral data to improve the performance of the algorithms. However, there is currently no standard for these methods, and no studies have compared them so far in the medical field. In this work, we evaluated different combinations of preprocessing steps, including spatial and spectral smoothing, Min-Max scaling, Standard Normal Variate normalization, and a median spatial smoothing technique, with the goal of improving tumor detection in three different HSI databases concerning colorectal, esophagogastric, and brain cancers. Two machine learning and deep learning models were used to perform the pixel-wise classification. The results showed that the choice of preprocessing method affects the performance of tumor identification. The method that showed slightly better results with respect to identifing colorectal tumors was Median Filter preprocessing (0.94 of area under the curve). On the other hand, esophagogastric and brain tumors were more accurately identified using Min-Max scaling preprocessing (0.93 and 0.92 of area under the curve, respectively). However, it is observed that the Median Filter method smooths sharp spectral features, resulting in high variability in the classification performance. Therefore, based on these results, obtained with different databases acquired by different HSI instrumentation, the most relevant preprocessing technique identified in this work is Min-Max scaling.

Joint Diagnostic Method of Tumor Tissue Based on Hyperspectral Spectral-Spatial Transfer Features

“…Liu found that the spectral data of nuclear compartments contribute more to the accurate staging of squamous cell carcinoma compared with peripheral regions [ 39 ]. Martinez-Vega evaluated different combinations of hyperspectral preprocessing steps in three HSI databases of colorectal, esophagogastric, and brain cancers [ 40 ], and he found that the choice of preprocessing method affects the performance of tumor identification, and this partly inspired our later data accumulation. However, problems involving the correlation between tumor tissues and typical spectral-spatial features [ 13 , 41 ], as well as the efficient transformation of conventional pathological and transfer spectral-spatial features [ 15 , 42 ], have not been effectively resolved.…”

Impact of Pre- and Post-Processing Steps for Supervised Classification of Colorectal Cancer in Hyperspectral Images

“…We calculated mean sensitivity and specificity using Inc_Norm_4_T for the same reduced patient dataset (12 patients) that was used in Martinez-Vega et al [ 26 ] and got sensitivity of 99.5% and specificity of 94% . The corresponding best values from Martinez-Vega et al were 86% and 87%.…”

“…That is why we will use the abbreviation “RS-based” for the second architecture. Another reason to use this architecture was that it has already shown good results on our data (Collins et al [ 25 ] and Martinez-Vega et al [ 26 ]). The main idea of this architecture is that it uses 3D convolutions followed by 1D convolutions.…”

“…SVMs have been evaluated to classify resectates of esophagogastric and colorectal cancers [ 25 ]. Convolutional neuronal networks (CNN) were used to classify ex vivo and in vivo colorectal cancer [ 25 ], esophagogastric cancer [ 25 ], brain [ 26 ], head and neck tumors [ 27 ], and thyroid and salivary tumors [ 28 ].…”

“…Solution of these problems can be found in the remote sensing (RS) area where HSI plays an important role in detecting objects. Several approaches to improve artificial networks were considered, such as testing different pre-processing steps (e.g., Standardization, Min-Max) [ 26 , 28 ], architectures (e.g., inception-based CNNs [ 28 ], DNN [ 29 ], also in combination with squeeze-and-excitation networks [ 30 ], and ResNet-based CNNs [ 31 ]), and post-processing [ 32 ].…”