Kerri-Ann Norton

Three-phase general border detection method for dermoscopy images using non-uniform illumination correction.

Computer‐aided diagnosis of dermoscopy images has shown great promise in developing a quantitative, objective way of classifying skin lesions. An important step in the classification process is lesion segmentation. Many studies have been successful in segmenting melanocytic skin lesions (MSLs), but few have focused on non‐melanocytic skin lesions (NoMSLs), as the wide variety of lesions makes accurate segmentation difficult.

Four-Class Classification of Skin Lesions With Task Decomposition Strategy

This paper proposes a new computer-aided method for the skin lesion classification applicable to both melanocytic skin lesions (MSLs) and nonmelanocytic skin lesions (NoMSLs). The computer-aided skin lesion classification has drawn attention as an aid for detection of skin cancers. Several researchers have developed methods to distinguish between melanoma and nevus, which are both categorized as MSL. However, most of these studies did not focus on NoMSLs such as basal cell carcinoma (BCC), the most common skin cancer and seborrheic keratosis (SK) despite their high incidence rates. It is preferable to deal with these NoMSLs as well as MSLs especially for the potential users who are not enough capable of diagnosing pigmented skin lesions on their own such as dermatologists in training and physicians with different expertise. We developed a new method to distinguish among melanomas, nevi, BCCs, and SKs. Our method calculates 828 candidate features grouped into three categories: color, subregion, and texture. We introduced two types of classification models: a layered model that uses a task decomposition strategy and flat models to serve as performance baselines. We tested our methods on 964 dermoscopy images: 105 melanomas, 692 nevi, 69 BCCs, and 98 SKs. The layered model outperformed the flat models, achieving detection rates of 90.48%, 82.51%, 82.61%, and 80.61% for melanomas, nevi, BCCs, and SKs, respectively. We also identified specific features effective for the classification task including irregularity of color distribution. The results show promise for enhancing the capability of the computer-aided skin lesion classification.

A 2D Mechanistic Model of Breast Ductal Carcinoma in Situ (DCIS) Morphology and Progression

Ductal carcinoma in situ (DCIS) of the breast is a non-invasive tumor in which cells proliferate abnormally, but remain confined within a duct. Although four distinguishable DCIS morphologies are recognized, the mechanisms that generate these different morphological classes remain unclear, and consequently the prognostic strength of DCIS classification is not strong. To improve the understanding of the relation between morphology and time course, we have developed a 2D in silico particle model of the growth of DCIS within a single breast duct. This model considers mechanical effects such as cellular adhesion and intra-ductal pressure, and biological features including proliferation, apoptosis, necrosis, and cell polarity. Using this model, we find that different regions of parameter space generate distinct morphological subtypes of DCIS, so elucidating the relation between morphology and time course. Furthermore, we find that tumors with similar architectures may in fact be produced through different mechanisms, and we propose future work to further disentangle the mechanisms involved in DCIS progression.

Classification of melanocytic skin lesions from non-melanocytic lesions

In this paper, we present a classification method of dermoscopy images between melanocytic skin lesions (MSLs) and non-melanocytic skin lesions (NoMSLs). The motivation of this research is to develop a pre-processor of an automated melanoma screening system. Since NoMSLs have a wide variety of shapes and their border is often ambiguous, we developed a new tumor area extraction algorithm to account for these difficulties. We confirmed that this algorithm is capable of handling different dermoscopy images not only those of NoMSLs but also MSLs as well. We determined the tumor area from the image using this new algorithm, calculated a total 428 features from each image, and built a linear classifier. We found only two image features, “the skewness of bright region in the tumor along its major axis” and “the difference between the average intensity in the peripheral part of the tumor and that in the normal skin area using the blue channel” were very efficient at classifying NoMSLs and MSLs. The detection accuracy of MSLs by our classifier using only the above mentioned image feature has a sensitivity of 98.0% and a specificity of 86.6% in a set of 107 non-melanocytic and 548 melanocytic dermoscopy images using a cross-validation test.

Skin lesion segmentation using an improved snake model

Accurate identification of lesion borders is an important task in the analysis of dermoscopy images since the extraction of skin lesion borders provides important cues for accurate diagnosis. Snakes have been used for segmenting a variety of medical imagery including dermoscopy, however, due to the compromise of internal and external energy forces they can lead to under- or over-segmentation problems. In this paper, we introduce a mean shift based gradient vector flow (GVF) snake algorithm that drives the internal/external energies towards the correct direction. The proposed segmentation method incorporates a mean shift operation within the standard GVF cost function. Experimental results on a large set of diverse dermoscopy images demonstrate that the presented method accurately determines skin lesion borders in dermoscopy images.

Development of a novel border detection method for melanocytic and non-melanocytic dermoscopy images

Computer aided diagnosis of dermoscopy images has shown great promise in developing a quantitative, objective way of classifying skin lesions. An important step in the classification process is the lesion segmentation. Many papers have been successful at segmenting melanocytic skin lesions (MSLs) but few have focused on non-melanocytic skin lesions (NoMSLs), since the wide variety of lesions makes accurate segmentation difficult. We developed an automatic segmentation program for the border detection of skin lesions. We tested our method on a set of 107 non-melanocytic lesions and on a set of 319 melanocytic lesions. Our method achieved precision/recall scores of 84.5% and 88.5% for NoMSLs, achieving higher scores than two previously published methods. Our method also achieved precision/recall scores of 93.9% and 93.8% for MSLs which was competitive or better than the two other methods. Therefore, we conclude that our approach is an accurate segmentation method for both melanocytic and non-melanocytic lesions.

Multiscale Modeling of Cancer

Breast cancer remains the second leading cause of cancer death in women, exceeded only by lung cancer. Specifically, triple-negative breast cancer (TNBC) has the worst prognosis, as it is more invasive and lacks estrogen, progesterone, and HER2 receptors that can be targeted with therapies. Due to the need for effective therapies for this type of breast cancer, it is critical to develop methods to (1) understand how TNBC progresses and (2) facilitate development of effective therapies. Here, we describe a multiscale model focusing on tumor formation. Our approach uses multiple scales to investigate the progression and possible treatments of tumors.

Emergence of Anti-Cancer Drug Resistance: Exploring the Importance of the Microenvironmental Niche via a Spatial Model

Practically, all chemotherapeutic agents lead to drug resistance. Clinically, it is a challenge to determine whether resistance arises prior to, or as a result of, cancer therapy. Further, a number of different intracellular and microenvironmental factors have been correlated with the emergence of drug resistance. With the goal of better understanding drug resistance and its connection with the tumor microenvironment, we have developed a hybrid discrete-continuous mathematical model. In this model, cancer cells described through a particle-spring approach respond to dynamically changing oxygen and DNA damaging drug concentrations described through partial differential equations. We thoroughly explored the behavior of our self-calibrated model under the following common conditions: a fixed layout of the vasculature, an identical initial configuration of cancer cells, the same mechanism of drug action, and one mechanism of cellular response to the drug. We considered one set of simulations in which drug resistance existed prior to the start of treatment, and another set in which drug resistance is acquired in response to treatment. This allows us to compare how both kinds of resistance influence the spatial and temporal dynamics of the developing tumor, and its clonal diversity. We show that both pre-existing and acquired resistance can give rise to three biologically distinct parameter regimes: successful tumor eradication, reduced effectiveness of drug during the course of treatment (resistance), and complete treatment failure. When a drug resistant tumor population forms from cells that acquire resistance, we find that the spatial component of our model (the microenvironment) has a significant impact on the transient and long-term tumor behavior. On the other hand, when a resistant tumor population forms from pre-existing resistant cells, the microenvironment only has a minimal transient impact on treatment response. Finally, we present evidence that the microenvironmental niches of low drug/sufficient oxygen and low drug/low oxygen play an important role in tumor cell survival and tumor expansion. This may play role in designing new therapeutic agents or new drug combination schedules.

Automated Reconstruction Algorithm for Identification of 3D Architectures of Cribriform Ductal Carcinoma In Situ

Ductal carcinoma in situ (DCIS) is a pre-invasive carcinoma of the breast that exhibits several distinct morphologies but the link between morphology and patient outcome is not clear. We hypothesize that different mechanisms of growth may still result in similar 2D morphologies, which may look different in 3D. To elucidate the connection between growth and 3D morphology, we reconstruct the 3D architecture of cribriform DCIS from resected patient material. We produce a fully automated algorithm that aligns, segments, and reconstructs 3D architectures from microscopy images of 2D serial sections from human specimens. The alignment algorithm is based on normalized cross correlation, the segmentation algorithm uses histogram equilization, Otsus thresholding, and morphology techniques to segment the duct and cribra. The reconstruction method combines these images in 3D. We show that two distinct 3D architectures are indeed found in samples whose 2D histological sections are similarly identified as cribriform DCIS. These differences in architecture support the hypothesis that luminal spaces may form due to different mechanisms, either isolated cell death or merging fronds, leading to the different architectures. We find that out of 15 samples, 6 were found to have ‘bubble-like’ cribra, 6 were found to have ‘tube-like’ criba and 3 were ‘unknown.’ We propose that the 3D architectures found, ‘bubbles’ and ‘tubes’, account for some of the heterogeneity of the disease and may be prognostic indicators of different patient outcomes.

MultiCellDS: a community-developed standard for curating microenvironment-dependent multicellular data

Exchanging and understanding scientific data and their context represents a significant barrier to advancing research, especially with respect to information siloing. Maintaining information provenance and providing data curation and quality control help overcome common concerns and barriers to the effective sharing of scientific data. To address these problems in and the unique challenges of multicellular systems, we assembled a panel composed of investigators from several disciplines to create the MultiCellular Data Standard (MultiCellDS) with a use-case driven development process. The standard includes (1) digital cell lines, which are analogous to traditional biological cell lines, to record metadata, cellular microenvironment, and cellular phenotype variables of a biological cell line, (2) digital snapshots to consistently record simulation, experimental, and clinical data for multicellular systems, and (3) collections that can logically group digital cell lines and snapshots. We have created a MultiCellular DataBase (MultiCellDB) to store digital snapshots and the 200+ digital cell lines we have generated. MultiCellDS, by having a fixed standard, enables discoverability, extensibility, maintainability, searchability, and sustainability of data, creating biological applicability and clinical utility that permits us to identify upcoming challenges to uplift biology and strategies and therapies for improving human health.