“…Among others, polarimetry allows the imaging of nerve fiber bundles in the human brain 9 and the inspection of Alzheimer disease 10 . Polarization-related techniques also demonstrate high accuracy in the early detection of some cancers such as skin cancer 11 , colon cancer 12 , breast cancer 13 , 14 and brain cancer 15 . Moreover, recent studies reveal the accurate performance of some machine learning algorithms built-in the polarimetric analysis of some biological tissues 16 – 20 .…”
Imaging polarimetry methods have proved their suitability to enhance the image contrast between tissues and structures in organic samples, or even to reveal structures hidden in regular intensity images. These methods are nowadays used in a wide range of biological applications, as for the early diagnosis of different pathologies. To include the discriminatory potential of different polarimetric observables in a single image, a suitable strategy reported in literature consists in associating different observables to different color channels, giving rise to pseudo-colored images helping the visualization of different tissues in samples. However, previous reported polarimetric based pseudo-colored images of tissues are mostly based on simple linear combinations of polarimetric observables whose weights are set ad-hoc, and thus, far from optimal approaches. In this framework, we propose the implementation of two pseudo-colored methods. One is based on the Euclidean distances of actual values of pixels and an average value taken over a given region of interest in the considered image. The second method is based on the likelihood for each pixel to belong to a given class. Such classes being defined on the basis of a statistical model that describes the statistical distribution of values of the pixels in the considered image. The methods are experimentally validated on four different biological samples, two of animal origin and two of vegetal origin. Results provide the potential of the methods to be applied in biomedical and botanical applications.
“…Among others, polarimetry allows the imaging of nerve fiber bundles in the human brain 9 and the inspection of Alzheimer disease 10 . Polarization-related techniques also demonstrate high accuracy in the early detection of some cancers such as skin cancer 11 , colon cancer 12 , breast cancer 13 , 14 and brain cancer 15 . Moreover, recent studies reveal the accurate performance of some machine learning algorithms built-in the polarimetric analysis of some biological tissues 16 – 20 .…”
Imaging polarimetry methods have proved their suitability to enhance the image contrast between tissues and structures in organic samples, or even to reveal structures hidden in regular intensity images. These methods are nowadays used in a wide range of biological applications, as for the early diagnosis of different pathologies. To include the discriminatory potential of different polarimetric observables in a single image, a suitable strategy reported in literature consists in associating different observables to different color channels, giving rise to pseudo-colored images helping the visualization of different tissues in samples. However, previous reported polarimetric based pseudo-colored images of tissues are mostly based on simple linear combinations of polarimetric observables whose weights are set ad-hoc, and thus, far from optimal approaches. In this framework, we propose the implementation of two pseudo-colored methods. One is based on the Euclidean distances of actual values of pixels and an average value taken over a given region of interest in the considered image. The second method is based on the likelihood for each pixel to belong to a given class. Such classes being defined on the basis of a statistical model that describes the statistical distribution of values of the pixels in the considered image. The methods are experimentally validated on four different biological samples, two of animal origin and two of vegetal origin. Results provide the potential of the methods to be applied in biomedical and botanical applications.
“…In the case of applications in biophotonics, polarimetric methods have proved to be very useful tools to enhance the image contrast of some organic structures, and/or providing information of certain structures invisible by using regular (non-polarimetric) images. This situation is useful, for instance, for the early detection of some diseases, such as breast cancer 7 , 8 , colon cancer 9 , skin cancer 10 , or brain cancer 11 , among others.…”
Section: Introductionmentioning
confidence: 99%
“…The above-stated use of polarimetric methods for the study of animal 12 or even human 8 – 10 tissues is a well-established field of work 13 and nowadays continues in constant development. However, the application of polarimetric methods for the study of plant diseases is less common and, in the last decade, there is a growing interest of exploring more complex (and rich in terms of information) polarimetric solutions for applications in plant science.…”
This paper highlights the potential of using polarimetric methods for the inspection of plant diseased tissues. We show how depolarizing observables are a suitable tool for the accurate discrimination between healthy and diseased tissues due to the pathogen infection of plant samples. The analysis is conducted on a set of different plant specimens showing various disease symptoms and infection stages. By means of a complete image Mueller polarimeter, we measure the experimental Mueller matrices of the samples, from which we calculate a set of metrics analyzing the depolarization content of the inspected leaves. From calculated metrics, we demonstrate, in a qualitative and quantitative way, how depolarizing information of vegetal tissues leads to the enhancement of image contrast between healthy and diseased tissues, as well as to the revelation of wounded regions which cannot be detected by means of regular visual inspections. Moreover, we also propose a pseudo-colored image method, based on the depolarizing metrics, capable to further enhance the visual image contrast between healthy and diseased regions in plants. The ability of proposed methods to characterize plant diseases (even at early stages of infection) may be of interest for preventing yield losses due to different plant pathogens.
“…Pham et al. 18 – 22 employed a Stokes–Mueller method to examine the polarization properties of skin cancer, liver cancer tissues, neuroblastoma, collagen-rich tendons, and cartilage. It was shown that the proposed method yielded nine effective parameters for distinguishing between normal skin tissue and various skin cancer tissues, including BCC, squamous cell carcinoma (SCC), and malignant melanoma.…”
Section: Introductionmentioning
confidence: 99%
“…Based on the fruitful achievement of Mueller matrix in Refs. 13 – 22 and machine learning techniques for skin cancer detection in Refs. 24 – 30 , furthermore, the RF classifier is adopted for this study because of its advantage for overcoming the overfitting and suitable for classifying untrained data.…”
.
Significance:
The Mueller matrix decomposition method is widely used for the analysis of biological samples. However, its presumed sequential appearance of the basic optical effects (e.g., dichroism, retardance, and depolarization) limits its accuracy and application.
Aim:
An approach is proposed for detecting and classifying human melanoma and non-melanoma skin cancer lesions based on the characteristics of the Mueller matrix elements and a random forest (RF) algorithm.
Approach:
In the proposal technique, 669 data points corresponding to the 16 elements of the Mueller matrices obtained from 32 tissue samples with squamous cell carcinoma (SCC), basal cell carcinoma (BCC), melanoma, and normal features are input into an RF classifier as predictors.
Results:
The results show that the proposed model yields an average precision of 93%. Furthermore, the classification results show that for biological tissues, the circular polarization properties (i.e., elements
,
,
, and
of the Mueller matrix) dominate the linear polarization properties (i.e., elements
,
,
, and
of the Mueller matrix) in determining the classification outcome of the trained classifier.
Conclusions:
Overall, our study provides a simple, accurate, and cost-effective solution for developing a technique for classification and diagnosis of human skin cancer.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.