2013.11 ~ 2014.01 TonguxDx: Stage3 Feasible Evalution
In the preliminary user experience, one of the authors (20s
female) used the TongueDx application and write body condition
diary for one month. the Figure actually shows the real
data about her tongue color graph. It shows in one day the
tongue body color significantly changed from Pale to Purple
and appeared unstable. In her health condition diary, it
referred that her digestive system was in trouble in that day
because of overeating and over stress, and felt uncomfortable
after that day. Therefore, we think our tongue color graph
could give users and doctors some useful reference value in
physical health inspection. As further feasibility test, we are
asking 20s collage students to use the TongueDx application.
動作を検証するために，筆者の一人（20 代女性）がTongueDx を1 ヶ月使用し，健康状況をノートに記録した．
図は，著者の実際の舌色変化グラフである．このグラフは，11 月9 日以降に舌の本体の色がPale からPurple へ変化していることを示している．このときの本人の健康記録によると，食べ過ぎとストレスのために消化器系不全に陥り，不快を訴えていたことが記されている．このように，舌色変化のグラフは，ユーザと医師の双方に対して，身体健康に関する有用な情報になると考えられる．現在，4 人の被験者（いずれも20 代女性）に依頼して，TongueDx アプリケーションの評価実験を進めている．
2013.04 ~ 2013.10 TonguxDx: Stage2 Development
On the TongueDx system, users can keep track of their health condition by recording the color of tongue coating and body on smartphones. In fact, our system uses tongue diagnosis techniques originated from Traditional Chinese Medicine (TCM) theories. In the theories, tongue symptom as one of the important diagnosing indicators can tell the health of human body. To avoid color error affected by surrounding light, a tongue color calibration by using teeth color is proposed to adjust white balance of the tongue picture. K-means algorithm is used to separate tongue coating from body. From the line graph of tongue coating and body color displayed on smartphones, people can know their health conditions timely. We have evaluated the TongueDx performance for one month in the preliminary user experience.
This paper is published in the Human Interface Society Symposium 2013.
Ryu Ini, Siio Itiro, TongueDx: A tongue diagnosis system for personal health care on smartphones, Interaction 2014, Tokyo, 2014.2 (accepted refereed demonstration) [acceptance rate:28.3%]
2012.10 ~ 2012.12 KausWalkTour
Internship in Centria laboratory, Finlnad as an ios developer.
KausWalkTour (or called trace to the masters' footprints) is an application applied to iPad which guides tourists' way to the sightseeings.
This project is made for Kaustinen - a folk music town in Finland. In this demo project, there are twelve pins on the map. Tourists can get iPad from Pelimannitalo (marked by number one) and give back it after visit all of the places.
Tourists' current location will be showed on the map by using GPRS. When they walk around the city followed by the pins marked one to twelve on the map, tourists can click the blue buttons on the left of screen and see some short descriptions or details information about the place they would be interested in. Moreover, when near the destination, it will be automatically open the view of texts, images and music about this place. The map can be rotate according to the north heading so that tourists can find their way easily.
This project was developed with two Netherlands guys. I am responsible for the development of iOS application.
2012.04 ~ 2012.07 TongueDx: Stage 1 Feasibility Study
Experiment 1 , The feasibility of color calibration by using teeth color as a standard.
In chinese traditional medical diagnosis, doctors observe the color and shape of tongue to analysis patient’s health condition. In this experiment, a feasible white balance calibration by teeth method is proposed to the tongue image color calibration. According to this experiment, a tongue diagnosis system applied to the mobile phone will be considered in further research.
Question: Is it possible to calibrate white balance of tongue image color by using teeth color as a standard?
Step 1: Take photos in different modes (see figure 1): Auto White Balance, Daylight, Tungsten, Fluorescent, Flash and underwater (camera Cannon G10 -> white balance settings).
Figure 1. Images in different modes.
Step 2: Pick random (n=5) pixels on teeth area, gray card area and pink color area - the closest to the tongue color. Then, calculate the average R.G.B values of random (n=5) pixels and use white balance algorithm (see figure 2.) to get calibrated pink's average R.G.B values.
Figure 2. One of the white balance algorithm.
In the case of Gray card as a standard:
Γ R = 129 / AveR;
Γ G = 129 / AveG;
Γ B = 129 / AveB.
In the case of Teeth color as a standard:
Γ R = AveR0 / AveR’;
Γ G = AveG0 / AveG’;
Γ B = AveB0 / AveB’.
*AveR,G,B is the average red, green, blue values of random pixels on the gray card area of raw image, AveR0,G0,B0 is on the teeth area of calibrated auto white balance image and AveR’,G’,B’ is on the teeth area of raw image.
Step 3: Compare the R.G.B error rates of Teeth with these of Raw (do nothing) and the Gray card according to the average R.G.B of random (n=5) pixels on pink color area in different modes, we get the line chart as followed :
Figure 3. The R.G.B error rates of Raw (do nothing), Gray card and Teeth in five different modes.
In figure 3, it shows two almost parallel error rate lines of Gray card and Teeth and one high error rate line of Raw (do nothing). It confirms the teeth color can be used as a standard to calibrate tongue image in tongue diagnosis system.