Dataset Information

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.

SUBMITTER: Martinez-Vega B

PROVIDER: S-EPMC9693077 | biostudies-literature | 2022 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

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

Martinez-Vega Beatriz B Tkachenko Mariia M Matkabi Marianne M Ortega Samuel S Fabelo Himar H Balea-Fernandez Francisco F La Salvia Marco M Torti Emanuele E Leporati Francesco F Callico Gustavo M GM Chalopin Claire C

Sensors (Basel, Switzerland) 20221118 22

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. Howe ...[more]

PMID: 36433516

Similar Datasets

Project description:BackgroundThe capacity of technologies measuring DNA methylation (DNAm) is rapidly evolving, as are the options for applicable bioinformatics methods. The most commonly used DNAm microarray, the Illumina Infinium HumanMethylation450 (450K array), has recently been replaced by the Illumina Infinium HumanMethylationEPIC (EPIC array), nearly doubling the number of targeted CpG sites. Given that a subset of 450K CpG sites is absent on the EPIC array and that several tools for both data normalization and analyses were developed on the 450K array, it is important to assess their utility when applied to EPIC array data. One of the most commonly used 450K tools is the pan-tissue epigenetic clock, a multivariate predictor of biological age based on DNAm at 353 CpG sites. Of these CpGs, 19 are missing from the EPIC array, thus raising the question of whether EPIC data can be used to accurately estimate DNAm age. We also investigated a 71-CpG epigenetic age predictor, referred to as the Hannum method, which lacks 6 probes on the EPIC array. To evaluate these epigenetic clocks in EPIC data properly, a prior assessment of the effects of data preprocessing methods on DNAm age is also required.MethodsDNAm was quantified, on both the 450K and EPIC platforms, from human primary monocytes derived from 172 individuals. We calculated DNAm age from raw, and three different preprocessed data forms to assess the effects of different processing methods on the DNAm age estimate. Using an additional cohort, we also investigated DNAm age of peripheral blood mononuclear cells, bronchoalveolar lavage, and bronchial brushing samples using the EPIC array.ResultsUsing monocyte-derived data from subjects on both the 450K and EPIC, we found that DNAm age was highly correlated across both raw and preprocessing methods (r > 0.91). Thus, the correlation between chronological age and the DNAm age estimate is largely unaffected by platform differences and normalization methods. However, we found that the choice of normalization method and measurement platform can lead to a systematic offset in the age estimate which in turn leads to an increase in the median error. Comparing the 450K and EPIC DNAm age estimates, we observed that the median absolute difference was 1.44-3.10 years across preprocessing methods.ConclusionsHere, we have provided evidence that the epigenetic clock is resistant to the lack of 19 CpG sites missing from the EPIC array as well as highlighted the importance of considering the technical variance of the epigenetic when interpreting group differences below the reported error. Furthermore, our study highlights the utility of epigenetic age acceleration measure, the residuals from a linear regression of DNAm age on chronological age, as the resulting values are robust with respect to normalization methods and measurement platforms.

Dataset Information

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

Publications

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

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets