1. Using FPGA to Provide Faster Digitally Enhanced Images in Order to Demonstrate a More Efficient Way to Process Images When Compared to Using Software.
James Haralambides • Darnell Henry • Jonathan Fineout
Background
Abstract
Embedded Medical System
Applications
The purpose of this research is to explain how beneficial FPGAs are to the medical field. With FPGA, data manipulation can be faster and more efficient in enhancing image quality for better use in diagnosing diseases.
This research can help individuals in the medical field achieve quicker and more accurate results. With FPGA, we will implement functions such as edge detection, de-skewing, compression, filtering, and other types of image manipulation through hardware rather than software. Edge detection can help a doctor more easily identify abnormalities present in an X-ray or CAT scan, or aid in the enhancement of an image while using a surgical microscope. Efficient image and data compression, used to save space in a medical database, is necessary when storing patient information and reduce costs. With filtering, different variations of an image can be processed to aid in other image distinctions.
To prove its usefulness, we will gather image data and perform various filters on it in order to enhance certain areas of the image. Such filters can be used to identify diseases and cancers found in patients. Although, seemingly simplistic, this is an extremely valuable capability. Programmable systems are already used in most medical equipment, and as technology improves, the programmable devices become more powerful and efficient, revealing a clearer need.
In the healthcare industry, the necessity to diagnose patients quickly and accurately is of utmost importance, but as always, cost is an important factor. It is fortunate that the use of programmable logic devices proves to be a cheaper option. Overall, FPGAs are essential in the healthcare industry.
Embedded Systems are found in a multitude of devices, from portable electronics such as cell phones and digital cameras, to large instruments capable of performing operations without the use of software.
The main objective of this project is to develop a medical system that is implemented through hardware rather than software. Allowing faster processing for image detection. With image detection built into hardware, it is possible for the same device to be used in a multitude of other image devices without the need of writing new software.
Using a Spartan 3E board, we will show just how efficient and accurate FPGA’s can process image data to enhance, filter, and manipulate a picture to provide better detection of objects. Image processing such as edge detection, de-skewing, compression, and other forms of filters will reveal more information present in a CAT scan or X-Ray.
Edge detection in a robotic system for minimally invasive surgery.
MRI reconstruction creates cross-sectional images of the human body.
Motion correction of X-ray images uses a dewarping function to sharpen results.
Image enhancement is commonly done with linear filtering. High-pass makes an image clearer but also increases noise. Low-pass blurs an image and decreases noise. Linear-combination filtering is a balance of these two, with output of an enhanced image with reduced noise.
Function
Description
Deinterlacer
Converts interlaced video formats to progressive video format
Color Space Converter
Converts image data between a variety of different color spaces
Scaler
Resizes and clips image frames
Alpha Blending Mixer
Mixes and blends multiple image streams
Gamma Correction
Performs gamma correction on a color plane/space
2D Filter
Implements 3x3, 5x5, or 7x7 finite impulse response (FIR) filter operation on an image-data stream to smooth or sharpen images
2D Median Filter
Implements a 3x3, 5x5, or 7x7 filter that removes noise in an image by replacing each pixel value with the median of neighboring pixel values
Line Buffer Compiler
Efficiently maps image line buffers to Altera on-chip memories
Implementation and Methods
Our project consists of using the Xilinx Spartan 3E board and a computer to export image data to the board for manipulation.
This board is capable of taking electrical signals and producing an output based on logical manipulation.
Our system will accept bits as input from image data and manipulate the bits to alter the image.
Once the image is received, various filters will be created by experimenting with different bit manipulations.
References
Knoll, Alois and Christob Staub “Image Processing for Medical Robotics.” Technische Universität München. 15 Mar <
Bohm, A.P.W., M. Chawathe, et al. “High Performance Image Processing on FPGAs” Colorado State University
Altera. “Medical Imaging Implementation Using FPGAs” Mar <