The number of research papers, where neural networks are applied in medical image analysis is growing. There is a proof that Convolutional Neural Networks (CNN) are able to differentiate skin cancer from nevi with greater accuracy than experienced specialists on average (sensitivity 82% and 73% accordingly).1 Team's latest research2 allows achieving even greater accuracy, by using specific narrow-band illumination. Nevertheless, the overall probability of early skin cancer detection depends on the availability of diagnostic tools. If screening tools will be available to a high number of general practices, the chance of disease detection will increase. The previous research3 shows that scalable cloud service is able to process a high number of users. After a certain number of users, the overall cost of the system, including cloud processing expenses and cost of high computational power portable device, might be higher if compared to an on-premises solution, where each device is capable of diagnosing without Internet access. It might be cheaper to equip devices with additional neural processing unit (NPU) and exclude cloud processing. Another option is to make screening available by using the newest smartphones that are equipped with NPU.4 The problem of using the NPU is that they are limited in storage space, accuracy, and features. Therefore, a full-size CNN model should be adapted and minimized to fit in a limited NPU. Research reviews existing CNN optimization methods and proposes the most accurate for skin cancer diagnostics. The paper evaluates CNN prediction losses when the model's elements’ precision is reduced from 32 bits to 8 and rounded to integer values.
This study presents autonomous system for microorganisms’ growth analysis in laboratory environment. As shown in previous research, laser speckle analysis allows detecting submicron changes of substrate with growing bacteria. By using neural networks for speckle analysis, it is possible to develop autonomous system, that can evaluate microorganisms’ growth by using cheap optics and electronics elements.
System includes embedded processing module, CMOS camera, 670nm laser diode and optionally WiFi module for connecting to external image storage system. Due to small size, system could be fully placed in laboratory incubator with constant humidity and temperature. By using laser diode, Petri dish with microorganisms’ substrate is illuminated with speckle pattern. Embedded camera and processing system obtain images and stores them for processing with neural network.
Neural network utilizes “3D ConvNets” architecture with ability to encode not only spatial speckle variance, but also their changes in time. Convolutive approach allows significantly reduce the number of trained parameters, therefore reducing training and detection time. Neural network training used 200 bacteria colonies and additional 300 areas without bacteria. In the result, trained neural network reaches 0.95 accuracy score, that proves correctness of the approach.
In this study 300 skin lesion (including 32 skin melanomas) multispectral data cubes were analyzed. The multi-step and single step machine learning approaches were analyzed to find the wavebands that provide the most information that helps discriminate skin melanoma from other benign pigmented lesions. The multi-step machine learning approach assumed training several models but proved itself to be ineffective. The reason for that is a necessity to train a segmentation model on a very small dataset and utilization of standard machine learning classifier which have shown poor classification performance. The single-step approach is based on a deep learning neural network. We have conducted 2600 experiments on two neural network architectures: popular pre-trained image analysis “InceptionV3” and simple custom convolutional neural network (ConvNet) classifiers. Observing performance metrics of these two deep-learning (DL) based architectures allowed to determine combinations of three spectral wavebands allowing to train a classifier with the best classification results. It was found that a simple ConvNet classifier allowed us to get better classification results. ConvNet training results have shown that most informative wavebands are 450nm which is the most informative for melanin concentration on the skin surface, 590nm that represents integral information about melanin and hemoglobin distribution from epidermis-dermis layer, and 950 nm that provide information from deeper skin layers. As introduced the convolutional neural network (CNN) model was simple but has not shown great performance. Also, we have to explore alternative CNN architectures. AutoKeras framework was used to find an architecture of the image classifier using the found waveband triplets.
Non-invasive skin cancer diagnostic methods develop rapidly thanks to Deep Learning and Convolutional Neural Networks (CNN). Currently, two types of diagnostics are popular: (a) using single image taken under white illumination and (b) using multiple images taken in narrow spectral bands. The first method is easier to implement, but it is limited in accuracy. The second method is more sensitive, because it is possible to use illumination considering the absorption bands of the skin chromophores and the optical properties of the skin. Currently CNN use a single white light image, due to the availability of large datasets with lesion images. Since CNN processing and analysis requires a large image database, only mathematical models have been used for multispectral diagnostics so far. Several scientific groups have created unique CNN, but the possibility of sharing pre-trained CNN models is limited due to the diversity of spectral bands used for skin lesion imaging. Current CNN models require image sets where each skin lesion has the same number of specific spectral bands. Therefore, researchers are unable to share their trained CNN models and each team uses a limited amount of skin lesions for CNN training. The paper proposes multi-input CNN architecture with a special encoding layer that allows using images in arbitrary spectral bands. That will allow sharing pre-trained models with other researchers to fine tune the model using additional wavebands. Additionally, the proposed model can adopt images taken under white illumination. As a result, it will be able to increase current melanoma detection accuracy
The paper proposes an approach of a novel non-contact optical technique for early evaluation of microbial activity. Noncontact evaluation will exploit laser speckle contrast imaging technique in combination with artificial neural network (ANN) based image processing. Microbial activity evaluation process will comprise acquisition of time variable laser speckle patterns in given sample, ANN based image processing and visualization of obtained results. The proposed technology will measure microbial activity (like growth speed) and implement these results for counting live microbes. It is expected, that proposed technology will help to evaluate number of colony forming units (CFU) and return results two to six times earlier in comparison with standard counting methods used for CFU enumeration.
The goal of our study is to train artificial neural networks (ANN) using multispectral images of melanoma. Since the number of multispectral images of melanomas is limited, we offer to synthesize them from multispectral images of benign skin lesions. We used the previously created melanoma diagnostic criterion p'. This criterion is calculated from multispectral images of skin lesions captured under 526nm, 663nm, and 964nm LED illumination. We synthesize these three images from multispectral images of nevus so that the p' map matches the melanoma criteria (the values in the lesion area is >1, respectively). Demonstrated results show that by transforming multispectral images of benign nevus is possible to get a reliable multispectral images of melanoma usable for ANN training.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.