How Intelligent Medical Imaging Technology can motivate Higher






![Database construction Sources : [5] Kugure, N. , Knowledge of Pet Dog Disease and Database construction Sources : [5] Kugure, N. , Knowledge of Pet Dog Disease and](https://slidetodoc.com/presentation_image_h2/79778a275a78854af6740fda4a94c02e/image-7.jpg)


























![Experiment Performance index : Locating Key Point Accurately [11] Kim, K. B. , Park, Experiment Performance index : Locating Key Point Accurately [11] Kim, K. B. , Park,](https://slidetodoc.com/presentation_image_h2/79778a275a78854af6740fda4a94c02e/image-34.jpg)








- Slides: 42

How Intelligent Medical Imaging Technology can motivate Higher Qualitative Healthy Life? Prof. Doo Heon Song Dept. of Computer Games, Yong-in Song. Dam College, Korea dsong@ysc. ac. kr Nov. 27, 2015 at 國立宜蘭大學

Machine Intelligence can guide Pet Dog Health Pre-Diagnosis for Casual Owner: A Neural Network Approach International Journal of Bio-Science and Bio. Technology 6, no. 2 (2014): 83 -90.

Motivation of the Study Pet Dog is now almost a family member. Pet-attached people may have emotional loss from pet dog's disease or sudden death. (Wieler LH et. als 2011 International Journal of Medical Microbiology) Increasing risk of transmission of microorganisms between humans and their dogs No reliable and COMPREHENSIBLE information about Pet’s diseases/symptoms. Pet gives sign…BUT…. . I can’t catch it.

My dog is SICK!

Veterinarian says “It’s too Late”

Necessary Conditions of Ad-hoc Diagnosis System 1. Reliable Standardized Disease-Symptoms Relationships as a Database form. 2. Learning Engine to decide what are “most probable” diseases with given symptoms. 3. Veterinarians should be involved. 4. Pet owner is NOT an medical Expert.
![Database construction Sources 5 Kugure N Knowledge of Pet Dog Disease and Database construction Sources : [5] Kugure, N. , Knowledge of Pet Dog Disease and](https://slidetodoc.com/presentation_image_h2/79778a275a78854af6740fda4a94c02e/image-7.jpg)
Database construction Sources : [5] Kugure, N. , Knowledge of Pet Dog Disease and Taming, Yuwon Print Ltd. 2001 [6] Dateuchi Y. , Pet Dog Diseases and Treatments, Haseo Print Ltd. , 2001

Database Tables Diseases ID 1 2. . 105 Disease Symptom Codes Description ascariasis trichuriasis. . 2 -3 -4 -5 -62 -70 1 -2 -6 -62 -70 -71 -80. . Expression . . . prostatomegaly 2 -70 -80 -171 Expression Symptoms ID 1 2. . . 180 Part Whole Body. . . Temper Expression. . Learning Result Symptom anemia inappetence. . . Hide in the dark Cluster Input Neuron Strength 1 1. . 1 2. . 0 1. . . 105 180 0

System Behavior

Why ART 2? 1) ART 2 is a self-organizing (no target value) pattern clustering structure by competitive learning. 2) It is a stable and adaptable neural network with incremental learning ability, that is, new learning procedure does not affect already existing clusters. 3) No local minima problem. 4) It is learnable with binary input and analogue input. 5) The change of connection strength is the average of all input patterns thus it is uniformly distributed to all clusters.

ART 2 Learning

Example UI

Sample Outputs galactorrhea(50%), corneitis(25%), external otitis(11%). Diagnosis with Secondary Symptom Cataract (72%) corneitis(25%) galactorrhea(23%).

Conclusion 1. The role of Ad-hoc Pre-diagnosis system is not replacing veterinarian's diagnosis but stimulating owner's attention to pet dog's abnormality. 2. Technically we need reliable database and inference engine with uncertainty – ART 2 does it for us. 3. Some diseases have different symptoms with respect to the progress of the disease. In those cases, veterinarian's care is "must". 4. How about CATS? – MUST have reliable disease-symptoms database first and yet to find it. 5. Mobile application can be developed.

Intelligent Automatic Extraction of Canine Cataract Object for Developing Handy Pre-Diagnostic Tool Info 2015, Taipei, Taiwan

Motivation of The Research . Cataract is a common disease for pet dogs developed with aging (>= 8 Yrs in general , Almost all > 13. 5 Yrs). Pet gives sign…BUT…. . I can’t catch it. Your dog would express more degree of attachment to the owners than usual and/or staggers in walk. While it’s a serious disorder, the treatment is easier if the symptom is found earlier by the owner. Need a Software for Pet Owners to see of the pet has Cataract!

Your Dog is suffering Cataract – Cloudy Vision Eyedrops Surgery

Image Analysis from Digital Camera Image from Non-Professional Digital Camera contains irregular pixel values. Ultrasonography, MRI , and endoscopic evaluation technique are all for professional Veterinarian Should extract the area of Cataract Automatically. A complete evaluation of eye by veterinary ophthalmologist will determine if the cataract treatment is necessary. Our Goal is to develop Handy Pre-Diagnosing Tool with No Deep Knowledge. Just alert them to watch the Pet!

Cataract Extraction Processes

Fuzzy Stretching. , (2. 2)

ART 2 Learning

Effect of ART 2 based Quantization Binarization

8 -Directional Contour Tracing.

Histogram Analysis and Cataract Extraction

Successful Cases

Failed Extraction The software can not discriminate white hair around eyes from cataract in ART 2 clustering process !

Automatic Extraction of Organs from Ultrasonography with Intelligent Imaging - Various Ongoing Researches…. .

Advantages of ultrasonography Non-invasive Real time examination Portable, quick application Excellent resolution of superficial structures Flow study without using contrast media Less expensive

Limitations of ultrasonography Media dependant Impossible overcome of the too high difference of sound impedance (Air, bone, calcification, metal etc) Problem of penetrating power and resolution in deep location Narrow view field Operator dependant Lack of objectivity

1. Cervical Vertebrae and Neck Pain is common and a serious threat of healthy life. Clinical neck pain is associated with impairment of muscle performance Kim, Kwang Baek, Doo Heon Song, Hyun Jun Park, and Sungshin Kim. "Automatic Extraction of Cervical Vertebrae from Ultrasonography with Fuzzy ART Clustering. " In Advances in Neural Networks–ISNN 2015, pp. 297 -304. Springer International Publishing, 2015.

Fig 1. (A) Ultrasound probe on the neck during set-up. (B) Region of interests (ROIs, each 10 mm long) as positioned in the longitudinal ultrasound image of the dorsal neck muscle Probing Cervical Vertebrae by Ultrasound (A) Ultrasound probe on the neck during set-up. (B) Region of interests (ROIs, each 10 mm long) as positioned in the longitudinal ultrasound image of the dorsal neck muscles from the most superficial to the deepest layer: trapezius, splenius, semispinalis capitis, semispinalis cervicis, and multifidus muscles From van der Werff, Ross, Shaun O'Leary, Gwendolen Jull, Michael Peolsson, Johan Trygg, and Anneli Peolsson. "A speckle tracking application of ultrasound to evaluate activity of multilayered cervical muscles. " Journal of rehabilitation medicine 46, no. 7 (2014): 662 -667.

Structure of Cervical Vertebrae in Ultrasonography DCF muscles have major roles in maintaining cervical lordosis and providing cervical joint stabilization. Sternocleidomastoid muscle (SCM) is related with the rotation of the neck

We Need Automatic Vision based Object Extractor/Analyzer 1. Ultrasonography analysis is non-invasive, inexpensive, and responding in real time. But, the performance is largely dependent on the inspector’s expertise. 2. Our concern is to detect and extract muscles such as SCM and DCFs in conjunction with cervical vertebrae automatically from ultrasonography and measuring the thickness for further medical analysis 3. Aim to locate measuring key point accurately to avoid manual subjective key point setting for muscle analysis. 4. Locating cervical vertebrae and related DCF are more important and difficult due to low brightness contrast among objects.
![Experiment Performance index Locating Key Point Accurately 11 Kim K B Park Experiment Performance index : Locating Key Point Accurately [11] Kim, K. B. , Park,](https://slidetodoc.com/presentation_image_h2/79778a275a78854af6740fda4a94c02e/image-34.jpg)
Experiment Performance index : Locating Key Point Accurately [11] Kim, K. B. , Park, H. J. , Song, D. H. , Han, S. S. Extraction of Sternocleidomastoid and Longus Capitis/Colli Muscle Using Cervical Vertebrae Ultrasound Images. Current Medical Imaging Reviews, 10(2), 95 -104 (2014)

Experiment Summary The proposed method is implemented with C++ under Microsoft Visual Studio 2010 on the IBM-compatible PC with Intel(R) Core(TM) i 7 -2600 CPU @ 3. 40 GHz and 4 GB RAM. The experiment uses fifty two 800 x 600 size DICOM format images. In order to avoid human subjectivity, our ground truths of measuring points are obtained by two physical therapists’ agreements.

Related Experiment – DCF Extraction DCF Thickness Error Magnitude < 0. 3 cm for most cases

2. Abdominal Muscle Analysis

Extraction of Muscles Kim, Kwang-Baek, Hae-Jung Lee, Doo Heon Song, and Young Woon Woo. "Extracting fascia and analysis of muscles from ultrasound images with FCM-based quantization technology. " Neural Network World 20, no. 3 (2010): 405.

3. Appendix Extraction Appendicitis, an inflammation of the appendix, is the most common abdominal surgical emergency Kim, Kwang Baek, Hyun Jun Park, Doo Heon Song, and Sang-suk Han. "Developing an Intelligent Automatic Appendix Extraction Method from Ultrasonography Based on Fuzzy ART and Image Processing. " Computational and Mathematical Methods in Medicine 2015 (2015).

Appendix Extractions

SCIE Special Issue Editing http: //www. hindawi. com/journals/cin/si/296897/cfp/

For further questions/discussions, please email to; dsong@ysc. ac. kr