TL;DR
This paper introduces a novel near-infrared (NIR) vein visualization device designed to improve venipuncture success rates, particularly in challenging cases like infants or individuals with darker skin tones. By employing a unique angled NIR light penetration method and advanced image processing, the device aims to overcome the limitations of existing technologies in visualizing deeper veins and thicker body parts.
What Problem Does This Paper Address?
Venipuncture, a common medical procedure for blood draws or donations, can be surprisingly difficult. Factors like subcutaneous fat, dark skin pigmentation, or the patient's age can make veins hard to see, forcing practitioners to rely on anatomical knowledge and experience, which can lead to errors and patient harm. While existing vein visualization devices using near-infrared (NIR) light exist, they often struggle with visualizing deeper veins or veins in thicker body parts like the arm. Furthermore, some devices are bulky, inconvenient, or prohibitively expensive. This paper aims to address these limitations by developing a more effective, portable, and potentially more affordable vein visualization device.
Methods
The proposed vein-visualizing device combines a novel NIR light module design with an advanced image processing algorithm.
NIR Light Module: * Angled Penetration: Unlike traditional penetration methods that shine light perpendicularly, this device uses two lines of NIR diodes that emit light at a specific angle. This "penetration effect" is designed to improve visualization in thicker tissues. * 850 nm Wavelength: The device utilizes an 850 nm band-pass filter and 850 nm emitting diodes. This wavelength is chosen because it offers a good balance between light penetration and absorption by blood, creating a strong contrast between veins and surrounding tissue. * Prototype Design: The NIR diode module was designed using SolidWorks and 3D printed, incorporating two lines of 20 NIR diodes. A NIR CCD camera captures the images.
Image Processing Algorithm: * Region of Interest (ROI) Setting: To focus processing and improve efficiency, a specific region of interest is identified. * Complex Histogram Equalization (CHE): This method combines General Histogram Equalization (GHE) and Contrast Limited Adaptive Histogram Equalization (CLAHE) to enhance contrast and reduce non-uniform lighting effects. * Frangi Filter: A 2D Frangi filter is applied to enhance vessel-like structures by analyzing the Hessian matrix of the image intensity. * Adaptive Thresholding: This technique locally computes threshold values to effectively segment vein structures, even in images with non-uniform lighting. * Morphology Closing: This operation (dilation followed by erosion) is used to fill small holes and connect broken segments in the vein pattern. * Skeletonization: The Zhang-Suen thinning algorithm is used to create a thin, centered line representation of the vein structure, aiding in accurate overlay onto the skin.
Datasets
Not discussed in the paper. The experiments primarily utilized a custom-made tissue-mimicking phantom and in vivo testing on human subjects.
Key Insights and Ideas
- Angled NIR Penetration is Key: The core innovation lies in the angled NIR light emission. This approach, determined through phantom studies, appears to be more effective than perpendicular methods for penetrating thicker tissues and visualizing deeper veins.
- Optimized Wavelength for Contrast: The choice of 850 nm NIR light is crucial. It leverages the "NIR window" where skin absorption is lower, while blood absorption remains high, creating a significant contrast for vein detection.
- Image Processing is Crucial for Clarity: Simply capturing NIR images isn't enough. The multi-stage image processing pipeline (CHE, Frangi filter, adaptive thresholding, skeletonization) is essential for transforming raw, noisy images into clear, actionable vein visualizations.
- Phantom Studies are Vital for Parameter Tuning: The paper highlights the importance of phantom studies for determining critical design parameters like the penetration angle and depth, which are then used to build and test the prototype.
Key Results
- Phantom Study Success: The prototype demonstrated the ability to visualize vein models within a tissue phantom more distinctly than the naked eye.
- In Vivo Efficacy: The device successfully visualized veins on a human arm in vivo, with veins appearing clearly in the raw images.
- Improved Visualization in Challenging Cases: The proposed device was able to visualize vein patterns in the cubital fossa of a patient with invisible veins, a feat not achieved by conventional reflection methods.
- Comparative Performance: In a comparison with commercial devices like Veinviewer Flex and Veinlite EMS, the prototype showed comparable or superior performance in visualizing veins at various depths, particularly at 15 mm depth where other devices struggled. The prototype was also noted to be more cost-effective.
Limitations and Future Work
- Rough Estimation of Penetration Parameters: The authors acknowledge that the estimation of penetration parameters (depth and angle) was based on a simplified model and did not fully account for light scattering and intensity decrease within the tissue. This could lead to errors and affect effectiveness.
- Undesired Skeleton Lines: The image processing algorithm sometimes produced extraneous skeleton lines that did not correspond to actual vein structures, potentially causing confusion.
- Future Work: The authors suggest addressing these limitations by improving the accuracy of penetration parameter estimation and refining the image processing to eliminate false skeleton lines. They also mention that using a tourniquet and cleaning the skin with an alcohol swab could further improve results by removing the thin oily layer. The simple configuration and potential for stereo camera integration also point towards future improvements in depth information and user-friendliness.
Summary updated