Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

CA : A Cancer Journal for Clinicians

Most of the signal processing that we will study in this course involves local operations on a signal, namely transforming the signal by applying linear combinations of values in the neighborhood of each sample point. You are familiar with such operations from Calculus, namely, taking derivatives and you are also familiar with this from optics namely blurring a signal. We will be looking at sampled signals only. Let's start with a few basic examples. Local difference Suppose we have a 1D image and we take the local difference of intensities, DI(x) = 1 2 (I(x + 1) − I(x − 1)) which give a discrete approximation to a partial derivative. (We compute this for each x in the image.) What is the effect of such a transformation? One key idea is that such a derivative would be useful for marking positions where the intensity changes. Such a change is called an edge. It is important to detect edges in images because they often mark locations at which object properties change. These can include changes in illumination along a surface due to a shadow boundary, or a material (pigment) change, or a change in depth as when one object ends and another begins. The computational problem of finding intensity edges in images is called edge detection. We could look for positions at which DI(x) has a large negative or positive value. Large positive values indicate an edge that goes from low to high intensity, and large negative values indicate an edge that goes from high to low intensity. Example Suppose the image consists of a single (slightly sloped) edge:

http://ieeexplore.ieee.org/iel5/5/31307/01456462.pdf

Linear systems

Learning Deconvolution Network for Semantic Segmentation

Photoacoustic imaging is a promising method for visualizing biological tissues with optical absorbers. This article provides an overview of the rapidly developing field of photoacoustic imaging. Photoacoustics, the physical basis of photoacoustic imaging, is analyzed briefly. The merits of photoacoustic technology, compared with optical imaging and ultrasonic imaging, are described. Various imaging techniques are also discussed, including scanning tomography, computed tomography and original detection of photoacoustic imaging. Finally, some biomedical applications of photoacoustic imaging are summarized.

Photoacoustic imaging in biomedicine

Intracranial hemorrhage is a critical conditional with the high mortality rate that is typically diagnosed based on head computer tomography (CT) images. Deep learning algorithms, in particular, convolution neural networks (CNN), are becoming the methodology of choice in medical image analysis for a variety of applications such as computer-aided diagnosis, and segmentation. In this study, we propose a fully automated deep learning framework which learns to detect brain hemorrhage based on cross sectional CT images. The dataset for this work consists of 40,367 3D head CT studies (over 1.5 million 2D images) acquired retrospectively over a decade from multiple radiology facilities at Geisinger Health System. The proposed algorithm first extracts features using 3D CNN and then detects brain hemorrhage using the logistic function as the last layer of the network. Finally, we created an ensemble of three different 3D CNN architectures to improve the classification accuracy. The area under the curve (AUC) of the receiver operator characteristic (ROC) curve of the ensemble of three architectures was 0.87. Their results are very promising considering the fact that the head CT studies were not controlled for slice thickness, scanner type, study protocol or any other settings. Moreover, the proposed algorithm reliably detected various types of hemorrhage within the skull. This work is one of the first applications of 3D CNN trained on a large dataset of cross sectional medical images for detection of a critical radiological condition

Deep 3D convolution neural network for CT brain hemorrhage classification

Objective: The objective of this study is to validate if ex-vivo multispectral photoacoustic (PA) imaging can differentiate between malignant prostate tissue, benign prostatic hyperplasia (BPH), and normal human prostate tissue. Materials and Methods: Institutional Review Board's approval was obtained for this study. A total of 30 patients undergoing prostatectomy for biopsy-confirmed prostate cancer were included in this study with informed consent. Multispectral PA imaging was performed on surgically excised prostate tissue and chromophore images that represent optical absorption of deoxyhemoglobin (dHb), oxyhemoglobin (HbO 2 ), lipid, and water were reconstructed. After the imaging procedure is completed, malignant prostate, BPH and normal prostate regions were marked by the genitourinary pathologist on histopathology slides and digital images of marked histopathology slides were obtained. The histopathology images were co-registered with chromophore images. Region of interest (ROI) corresponding to malignant prostate, BPH and normal prostate were defined on the chromophore images. Pixel values within each ROI were then averaged to determine mean intensities of dHb, HbO 2 , lipid, and water. Results: Our preliminary results show that there is statistically significant difference in mean intensity of dHb ( P P = 0.0251) between malignant prostate and normal prostate tissue. There was difference in mean intensity of dHb ( P Conclusion: Our preliminary results of ex-vivo human prostate study suggest that multispectral PA imaging can differentiate between malignant prostate, BPH and normal prostate tissue.

Multispectral Photoacoustic Imaging of Prostate Cancer: Preliminary Ex-vivo Results.

OBJECTIVE. The purpose of this study was to validate whether ex vivo multispectral photoacoustic imaging can be used to differentiate malignant tissue, benign nodules, and normal human thyroid tissue. SUBJECTS AND METHODS. Fifty patients undergoing thyroidectomy because of thyroid lesions participated in this study. Multispectral photoacoustic imaging was performed on surgically excised thyroid tissue, and chromophore images that represented optical absorption of deoxyhemoglobin, oxyhemoglobin, lipid, and water were reconstructed. After the imaging procedure, the pathologist marked malignant tissue, benign nodules, and normal regions on histopathologic slides, and digital images of the marked histopathologic slides were obtained. The histopathologic images were coregistered with chromophore images. Areas corresponding to malignant tissue, benign nodules, and normal tissue were defined on the chromophore images. Pixel values within each area were averaged to determine the mean intensities of deoxyhemoglobi...

https://www.ajronline.org/doi/pdf/10.2214/AJR.13.11433

Preliminary Results of Ex Vivo Multispectral Photoacoustic Imaging in the Management of Thyroid Cancer

Basics and Clinical Applications of Photoacoustic Imaging

Objectives
The purpose of this study was to investigate the feasibility of differentiating malignant prostate from benign prostatic hyperplasia (BPH) and normal prostate tissue by performing frequency domain analysis of photoacoustic images acquired at 2 different wavelengths.

Methods
We performed multiwavelength photoacoustic imaging on freshly excised human prostate specimens taken from a total of 30 patients undergoing prostatectomy for biopsy-confirmed prostate cancer. Histologic slides marked by a genitourinary pathologist were used as ground truth to define regions of interest (ROIs) in the photoacoustic images. Primarily, 3 different prostate tissue categories, namely malignant, BPH, and normal, were considered, while a fourth category named nonmalignant was formed by combining the ROIs corresponding to BPH and normal tissue together. We extracted 3 spectral parameters, namely slope, midband fit, and intercept, from power spectra of the radiofrequency photoacoustic signals corresponding to the 3 primary tissue categories.

Results
We analyzed data from 53 ROIs selected from the photoacoustic images of 30 patients. According to the histopathologic analysis, 19 ROIs were malignant, 8 were BPH, and 26 were normal. All the 3 spectral parameters and C-scan grayscale photoacoustic image pixel values were found to be significantly different (P < .01) between malignant and nonmalignant prostate as well as malignant and normal prostate.

Conclusions
Preliminary results of our ex vivo human prostate study suggest that spectral parameters obtained by performing frequency domain analysis of photoacoustic signals can be used to differentiate between malignant and nonmalignant prostate.

/pdf/evaluation-of-frequency-domain-analysis-of-a-multiwavelength-1cjn3upycm.pdf

Evaluation of Frequency Domain Analysis of a Multiwavelength Photoacoustic Signal for Differentiating Malignant From Benign and Normal Prostates Ex Vivo Study With Human Prostates

Navalgund Rao

Papers

Deep 3D convolution neural network for CT brain hemorrhage classification

Multispectral Photoacoustic Imaging of Prostate Cancer: Preliminary Ex-vivo Results.

Preliminary Results of Ex Vivo Multispectral Photoacoustic Imaging in the Management of Thyroid Cancer

Basics and Clinical Applications of Photoacoustic Imaging

Evaluation of Frequency Domain Analysis of a Multiwavelength Photoacoustic Signal for Differentiating Malignant From Benign and Normal Prostates Ex Vivo Study With Human Prostates