Handheld Melanoma Detector
Melanoma is the eighth most common cancer among men and women, with over 60,000 new cases and 8,000 deaths reported annually in the US alone. Catching it early is vital, because it is difficult and often impossible to treat in its later stages. Diagnosis is currently based mainly on the skill of the dermatologist in visually identifying, though the naked eye, a pigmented lesion as potentially melanoma or its precancerous state, followed by biopsy. As a result, many unnecessary biopsies are performed.
Biopsy takes time. Tissue needs to be removed from the patient, stored, delivered to a commercial or hospital lab, and evaluated to ascertain the necessity of further resection. Technologies and approaches used or being developed to assist clinicians conduct naked-eye visual exam include total body photography, dermoscopy, and in vivo reflectance confocal microscopy. These techniques help ascertain the necessity of a biopsy but do not themselves provide diagnostic information.
Raman spectroscopy is a well established technique for the in-vivo study biological tissue and diagnosing disease. It is a non-contact, non-destructive technique that can detect changes in chemical composition and/or molecular structure between healthy and diseased tissue. MEP intends to complete the development of a compact, portable Raman scanner allowing an area of skin containing suspicious tissue to be diagnosed in real time. This revolutionary system will be a valuable tool for ambulatory clinical evaluation and decision-making thus improving patient healthcare, and significantly reducing healthcare costs for everyone.
Building upon work already conducted by MEP members for other cancers (see e.g. Fig. 1), MEP will evaluate the Raman spectra of excised benign and dysplastic melanocytic nevi and compare them with lesions of melanoma, at excitation wavelengths between 550 nm and 1000 nm in order to develop a fundamental model of the spectrum of benign dysplastic and malignant melanocytic nevi. Our multi-wavelength Raman spectroscopy technique allows a detailed reconstruction of the cellular/molecular structures as well as cellular and intermolecular interactions in healthy and diseased tissues.
From the spectral data, we will determine the optimum wavelength of light that will distinguish between diseased and healthy tissue and use this to develop a real-time diagnostic system with a built-in scanning Raman probe.
We have succeeded in showing differences between pre-cancerous cells and normal tissue using our custom Raman spectrometer for analysis of both animal and human cancers. We have investigated primary cancer samples from brain, pancreas and breast and metastases to the bone. Our Raman data for breast cancer illustrate clearly observable differences between peak profiles for normal, benign, and cancerous tissues. Similar distinctions are apparent for brain cancers in which there is a great need for precise, real-time detection of the tumor margins during surgery.
The technology can be applied to more than one type of tumor. Since 2005, MEP members have collected data from 260 patients (including >100 brain tissue samples from 36 patients). These samples are from a variety of locations, tumors, and grades and often contain normal tissue at the margins. In order to develop a quantitative method to separate normal and cancerous human tissue spectra, we have developed a Discriminant Function Analysis (DFA) program that automatically identifies and discriminates the Raman peaks for classification into normal, benign and cancerous. Preliminary results for DFA analysis for human breast and brain tissues are shown in Figure 2. These tissues were classified with greater than 95% accuracy with respect to the identification by a histopathologist. This demonstrates the clinical significance of Raman, however a detailed analysis of cancer cells, their growth, and their microenvironment is critical to a true understanding of the cancer growth process.
Figure 2: Discriminant Function Analysis of human breast tissue (left) and brain tissue (right)
In another study we looked at the Raman spectra of pediatric renal tumors. Again, the Raman spectra showed clear differences between normal kidney, Wilms’ Tumor, and other pathologies of the kidney. It was also shown that frozen tissues have similar Raman spectra to their fresh counterparts. This should allow the development of a database of Raman spectra from readily available banked frozen specimens, rather than obtaining fresh specimens from the operating room. This is especially significant for rare cancers.
Previous research shows that Raman spectroscopy offers an opportunity for the development of portable, high resolution diagnostic tool for rapid identification of skin malignancy in clinical practice. Raman spectroscopy has validity as a method to differentiate between skin cancers and normal tissue.
Aim 1: Generate multispectral data of clinically benign, dysplastic, and malignant melanocytic nevi obtained by excisional biopsy at excitation wavelengths between 550 nm to 1000 nm to identify discrete peaks and variations in spectra that differentiate normal from diseased tissue. From the result of multispectral screening, the optimal wavelength of light that can best resolve Raman peaks from background interference will be determine and used in designing and developing the optical component of the prototype.
Aim 2: Design and develop a prototype Raman spectrometer with built-in scanning probe for clinical use. The conceptual design allows an area of skin consisting of healthy and suspicious tissue to be scanned and the Raman spectra obtained point by point. Raman measurements of the healthy skin provide background spectra. This baseline data accounts for skins variances between individual people thus providing automatic calibration of the device.
Aim 3: In-vitro prototype testing will be done for statistical validation. The results will be compared to those obtained by our laboratory multispectral system and the gold-standard pathology diagnosis.
We conceive a prototype design of a clinical Raman spectrometer with build-in scanning probe. We anticipate design improvements/alterations as we go through the prototype development process.
Generate multispectral data: As the first step of this project, we generate a multispectral Raman database. The database consists of spectra of a specified number of samples of excised benign and dysplastic melanocytic nevi and melanoma obtained at a multitude of excitation wavelengths. From this, we determine an optimal discrete wavelength for excitation as well as define a select number of spectral peaks useful in identification and differentiation between healthy and diseased tissues. Based upon the results of the multispectral screening, we define the optical parameters necessary to construct the prototype. During the first 12 months of this project we obtain the multispectral data as well as simultaneously, develop the software and instrument the prototype with components that are not wavelength dependent, such as the scanning mirrors and electronics for device control.
Prototype development: We anticipate the possibility of redesign and fabrication of components of the prototype as we proceed with prototype development. This is feasible since we have access to scientists, engineers, and fabrication experts who work in partnership, as well as to a state of the art clean room facility. We anticipate prototype development to take approximately 21 months.
As part of prototype development, the wavelength dispersive element will be built, as illustrated in Table 1, via several tasks: computer simulation, fabrication, and independent functionality testing. We perform computer simulation to test the feasibility of the initial design, redesign if necessary, then proceed to fabrication and functionality testing. Based upon the results of functionality testing this process may be repeated. We estimate 6 months for a working prototype of the optical wavelength dispersive element, and an additional 3 months for potential refinement.
After the dispersive element is complete, the prototype of the Raman spectrometer with built-in scanning probe will be constructed and evaluated against a known standard. The results will be compared to those obtained with our benchtop laboratory system. At this step we will be able to ascertain the overall resolution and sensitivity of the instrument and make modification/adjustments if necessary. We estimate 6 months for this process providing an additional 3 months for redesign and refinement of the device.
After finally assembly, the prototype will be tested on clinically dysplastic melanocytic nevi obtained by excisional biopsy and compare the results with those already obtained with our laboratory scale system. We estimate 6 months for this process. The next phase of this work is to evaluate the prototype in a clinical study. Future work will include in-vivobenign, dysplastic and neoplastic changes in human melanocytic lesionsprior to excisional biopsy. Post hocconcordance will be done by established light microscopic, immunohistochemical, proteomic and genomic diagnosis.