How to assess scar hypertrophy--a comparison of subjective scales and Spectrocutometry: a new objective method.

Scar hypertrophy is a significant clinical problem involving both linear scars from elective surgery and scars caused by trauma or burns. The treatment of hypertrophic scars is often time consuming, and patients may need to be followed up for months or even years. The methods for reliable quantification of scar hypertrophy are at present unsatisfying. We have developed a new, objective method, Spectrocutometry, for documentation and quantification of scar hypertrophy. The instrument is based on standardized digital imaging and spectral modeling and calculates the estimated concentration change of hemoglobin and melanin from the entire scar and also provides standardized images for documentation. Three plastic surgeons have assessed 37 scars from melanoma surgery using Spectrocutometry, the Vancouver scar scale, and the patient and observer scar assessment scale. The intraclass correlation coefficient for the Vancouver scar scale and the patient and observer scar assessment scale was lower than required for reliable assessment (r=0.66 and 0.60, respectively). The intraclass correlation coefficient for Spectrocutometry was high (r=0.89 and 0.88). A Bayesian network analysis revealed a strong dependency between the estimated concentration change of hemoglobin and scar pain. Spectrocutometry is a feasible method for measuring scar hypertrophy. It is shown to be more reliable than subjective rating in assessing linear surgical scars.

Lower extremity ulcer image segmentation of visual and near-infrared imagery.

BACKGROUND/PURPOSE: We propose an automatic ulcer segmentation system with a simple manual correction possibility. In addition to visual color information, we use near-infrared (NIR) images because NIR can penetrate deeper into tissue than visual light. The system is able to measure the surface area of a lower extremity ulcer segmented at its different stages and constructs corresponding healing curves over time. This knowledge is useful in monitoring lower extremity ulcers and helps clinicians select the most efficient therapy. METHODS: Eighteen lower extremity ulcers and one ulcer on the back were examined from 17 patients. The patients were elderly individuals residing in the long-term care department of the Vaasa city hospital. One of the patients (P14) had been diagnosed with diabetes. The inclusion criteria for patients were an ulcer with a suitable size for the imaging device and the free will to volunteer. We developed a four-band spectral digital camera to image the reflectance of the skin. We use the spectral image pixels, in visual light and NIR, in analysis of lower extremity ulcers. For segmentation, the support vector classifier was found to be the best one. The segmentation system is designed to analyze three main ulcer tissue classes: black/necrotic, yellow/fibrous and red/granulation tissue. RESULTS: The experiments conducted confirm the feasibility of our approach. In most cases, the computed healing curves correspond to those made manually. The maximum error rate of ulcer area measurement for red/granulation tissue is 33% for 20 cases. This corresponds to the results published in the literature. The black/necrotic tissue may be located deeper under the skin surface; hence, the ulcer boundaries are not well defined, allowing only a rough estimate, yielding a maximum error of 44% for the three cases analyzed. For yellow/fibrous tissue, we had only one image in our database, whose error value is 23%. CONCLUSION: We propose a new imaging system for segmentation and measurement of different kinds of ulcers. This system is useful in practice for analysis and measurement of ulcer surface areas and observation of their change over time, which helps clinicians in the treatment of ulcers.

Kernel-based spectral color image segmentation.

In this work, we propose a new algorithm for spectral color image segmentation based on the use of a kernel matrix. A cost function for spectral kernel clustering is introduced to measure the correlation between clusters. An efficient multiscale method is presented for accelerating spectral color image segmentation. The multiscale strategy uses the lattice geometry of images to construct an image pyramid whose hierarchy provides a framework for rapidly estimating eigenvectors of normalized kernel matrices. To prevent the boundaries from deteriorating, the image size on the top level of the pyramid is generally required to be around 75 x 75, where the eigenvectors of normalized kernel matrices would be approximately solved by the Nyström method. Within this hierarchical structure, the coarse solution is increasingly propagated to finer levels and is refined using subspace iteration. In addition, to make full use of the abundant color information contained in spectral color images, we propose using spectrum extension to incorporate the geometric features of spectra into similarity measures. Experimental results have shown that the proposed method can perform significantly well in spectral color image segmentation as well as speed up the approximation of the eigenvectors of normalized kernel matrices.