J Med Syst (2016) 40: 166

DOI 10.1007/s10916-016-0523-4

TRANSACTIONAL PROCESSING SYSTEMS

Neonatal Jaundice Detection System

Mustafa Aydın 1 & Fırat Hardalaç 2 & Berkan Ural 2 & Serhat Karap 2

Received: 26 January 2016 / Accepted: 9 May 2016 / Published online: 26 May 2016
# Springer Science+Business Media New York 2016

Abstract Neonatal jaundice is a common condition that oc- consisted with bilirubin results which are obtained from the
curs in newborn infants in the first week of life. Today, tech- standard blood test and the compliance rate is 85 %.
niques used for detection are required blood samples and other
clinical testing with special equipment. The aim of this study Keywords Neonatal jaundice . Bilirubin . Image processing .
is creating a non-invasive system to control and to detect the Image segmentation . Machine learning regressions
jaundice periodically and helping doctors for early diagnosis.
In this work, first, a patient group which is consisted from
jaundiced babies and a control group which is consisted from Introduction
healthy babies are prepared, then between 24 and 48 h after
birth, 40 jaundiced and 40 healthy newborns are chosen. Neonatal jaundice or neonatal icterus is a disease which has
Second, advanced image processing techniques are used on been known since ancient times and today, diagnostic and
the images which are taken with a standard smartphone and therapeutic methods are constantly being developed [1]. The
the color calibration card. Segmentation, pixel similarity and first information about neonatal jaundice was written in the
white balancing methods are used as image processing tech- book of BEin Regiment der Kinder^ by Metlinger in 1473 [2].
niques and RGB values and pixels’ important information are Neonatal jaundice or neonatal hyperbilirubinemia is clearly
obtained exactly. Third, during feature extraction stage, with seen when bilirubin level exceeds 5 mg/dl (85 mmol/l) in
using colormap transformations and feature calculation, com- infants’ blood [3]. In the first week of life, bilirubin level rises
parisons are done in RGB plane between color change values in every newborn infants. According to the worldwide stan-
and the 8-color calibration card which is specially designed. dards, jaundice is detected in almost 60 % of the healthy full-
Finally, in the bilirubin level estimation stage, kNN and SVR term babies and 80 % of the preterm babies [4].
machine learning regressions are used on the dataset which are During the fetal period, red blood cells in the baby’s blood
obtained from feature extraction. At the end of the process, are quite different than a normal human’s red blood cells. This
when the control group is based on for comparisons, jaundice red blood cells contain Bfetal hemoglobin (HbF)^ [5]. The
is succesfully detected for 40 jaundiced infants and the suc- most important feature of these red blood cells is rapidly
cess rate is 85 %. Obtained bilirubin estimation results are disintegrating after birth. In the body, because of this occasion,
new red blood cells which contain a new kind of hemoglobin
Badult hemoglobin (HbA)^ are created simultaneously [6]. In
This article is part of the Topical Collection on Transactional Processing newborn infants, two factors are responsible from develop-
Systems
ment of jaundice - the breakdown of HbF as it is replaced with
HbA, and the relatively immature metabolic pathways of the
* Fırat Hardalaç
[email protected] liver, which are unable to conjugate and so excrete bilirubin as
quickly as an adult [7]. Generally, bilirubin is processed in the
liver in the body and more of them is converted and sent to the
1
Pediatrics-Neonatology, Fırat University, Elazig, Turkey rectum [8]. When more of the bilirubin can not be processed
2
Electrical Electronics Engineering, Gazi University, Ankara, Turkey by baby’s liver, this bilirubin is become to accumulate in
166 Page 2 of 11 J Med Syst (2016) 40: 166

baby’s blood, which called as hyperbilirubinemia. Bilirubin is when the control group was based on, for 40 jaundiced new-
a yellow color pigment and, therefore, when excess of the borns, jaundice was detected successfully and the success rate
bilirubin is accumulated in the skin, symptoms of jaundice was 85 %. Obtained bilirubin estimation results were
occurs [9]. consisted with bilirubin results which were obtained from
Neonatal jaundice is a clinical condition which may resolve the standard blood test and the compliance rate was 85 %.
without any intervention in the majority of babies; however, it In literature, there were limited studies about jaundice de-
is highly important that this critical period should be passed tection for neonatals and there was not any adequate resources
under a doctor’s supervision [10]. Otherwise, when the serum about these works. So, more comments could not be made
bilirubin level exceeds the critical level; permanent brain dam- about their performance and achievements. Apart from this
age due to accumulation of the bilirubin in the brain, a disease study, there have been no works that contained advanced im-
called kernicterus, can clearly be seen [11, 12]. age processing algorithms and machine learning regressions
Severe neonatal jaundice is a pediatric emergency because yet. In the Results and Discussion part, comparisons and per-
it may cause to kernicterus which can result in chronic formance analysis were clearly done between the neonatal
handicapping conditions like sensorineural deafness or cere- jaundice system and some works that were related to the gen-
bral palsy in those who survive. The goal of the management eral jaundice. The statistical test results were obtained from
of severe neonatal jaundice is therefore to rapidly reduce the these limited works and neonatal jaundice system’s statistical
serum bilirubin level to prevent kernicterus. There are a vari- results were compared with the other works’ results in this part
ety of treatment methods about neonatal jaundice; which are in detail.
still being developed. Specific treatment is not necessary for
the low level of bilirubin in physiological jaundice; but pho-
totherapy is started if the total bilirubin level exceeds a reliable
point, usually by >15 mg/dl, with respect to the postnatal age Materials and methods
and weight of the patient, [13]. By using the phototheraphy
method, bilirubin is excreted from the body by urine through Data collection phase
the kidneys after converted to the water-soluable products
[14]. Another important treatment method of severe neonatal Methods that were used in this study were applied to 40
hyperbilirubinemia is the exchange transfusion [15]. For this healthy and 40 jaundiced babies’ images. For this work, the
procedure, when serum bilirubin level exceeds the critical necessary document of the ethical rules was taken by Fırat
levels, usually by >20–25 mg/dl, with respect to the clinical Hardalaç in Ankara, Turkey. Indeed, these data were
condition and postnatal age and weight of the patient, the only collected from Fırat University Faculty of Medicine
way is an exchange transfusion [16]. Neonatal Department with the cooperation of Assoc. Prof.
In this work, creating a novel and a non-invasive system for Mustafa Aydın. Test subjects’ age (hour), bilirubin level
detecting jaundice in newborns optimally was generally (mg/dl) and hemolysis (n,%) average values were 33 h,
aimed. First, the 8-color calibration card was used for elimi- 11.9 mg/dl and 15 %, respectively. Figure 1 illustrated the
nating the enviromental lightning problems and doing the
necessary comparisons with a high performance for the other
stages. This card was placed on the baby’s abdomen and pic-
tures of the baby were taken with a standard smartphone cam-
era in the proper position and these data were transferred to
MATLAB® environment to be processed. Second, obtained
images were analyzed with the color based image processing
techniques to detect the optimal color tone with using the
baby’s important body parts, then with using morphological
techniques, edits and improvements have been made. After
this step, data analysis and calculation were done by machine
learning regressions. The last step of the system is to deter-
mine and to estimate the optimal bilirubin level in the baby’
blood. In the experimental part of the work, a patient group
which was consisted from jaundiced babies and a control
group which was consisted from healty babies were prepared,
then between 24 and 48 h after birth, 40 jaundiced and 40
healthy newborns were chosen for these groups and analysis Fig. 1 The distribution of bilirubin levels versus age value (hours) for 40
and comparisons were generally done for 80 infants. Finally, jaundiced babies
J Med Syst (2016) 40: 166 Page 3 of 11 166

distribution of bilirubin levels versus age value (hours) for 40 SVR (Support Vector Regression) algorithms were used on
jaundiced babies. the dataset which were obtained from the second stage. The
During the data entry phase, while baby was in the recum- fourth stage of the system was bilirubin level estimation. At
bent position, the 8-color calibration card was placed on the this stage, bilirubin level estimation process was achieved
baby’s abdominal region. With using 12 MP Samsung Galaxy with using obtained regression results [21]. The details of
Alpha smartphone and when using this smartphone camera the processes were described below in detail.
with no flash, baby’s whole body photos were taken and re-
sults were obtained. The resolution value of the obtained im- Color based balancing
ages was 96 dpi and for the other stage, these images were
transferred to MATLAB® environment and image processing Image segmentation Segmentation was the first step of the
methods and machine learning regressions were used on these system. Image segmentation was described as seperating the
images, respectively. image in each of the regions where different properties were
kept. Color calibration card which was used in this work was
Structure of the system specially designed and to obtain this card an advanced laser
printer and an uncoated paper were used. The example of the
The flowchart of the system was generally given in Fig. 2 and color calibration card was given in Fig. 3 [21].
every step of the process was mentioned below in detail.
The type and the size of the image data were uint8 and 440 With using the color calibration card that was given in Fig. 3,
um × 440 um, respectively. The first stage of the system was segmentation process was performed on the babies’ images
called as color based balancing. At this stage, segmentation (Fig. 4) [21]. During segmentation, each color on the calibra-
process was performed on the images. With using segmenta- tion card was evaluated seperately and estimation process was
tion, important regions and their information were preserved achieved. The color calibration card made the baby indepen-
and also unused regions were rearranged in the black color dent from external stimuli and after using the card, the image
tone [17]. Second, with using pixel similarity method, some of was ready for processing [21]. The captured image was sub-
the missing colors on the color calibration card is provided to jected to the segmentation process. First, for this process, the
be brought back [18]. Third, with using white balancing meth- captured image was divided into the desired segments, then 8
od, images were being independent from ambient light, reflec- main colors and jaundiced regions’ colors were compared
tions and shadows [19]. The second stage of the system was [22]. Second, jaundiced regions were concretized via segmen-
called feature extraction. At this stage, the information data tation and other regions were eliminated. For segmentation
about baby’s skin and the colors of the color calibration card process, point-based and region-based detection techniques
were obtained in RGB, YCbCr and Lab color spaces with were used, respectively [23].
using colormap transformations and feature calculation [20]. After this step, third, the image was filtered through the
The third stage of the system was called machine learning Gauss filter. After this situation, environmental noises and
regressions. At this stage, kNN (k-Nearest Neighbor) and blurring were eliminated [24].

Fig. 2 Flowchart of the neonatal

jaundice detection system
166 Page 4 of 11 J Med Syst (2016) 40: 166

White balancing One of the important problems of the image

processing is that different results can be obtained for an im-
age because of environmental and lightning differences [29].
This occasion is usually undesirable. For an image, the same
results can be expected to be obtained from different light
intensities and this is generally desirable [30].
The aim of white balancing is to make the original image
independent from the luminous intensity [31]. According to
the luminous intensity and the location of the resource, some
differences can be obtained from the process of determining
the skin color. To eliminate this problem, normalized red,
Fig. 3 Color calibration card example [21] green and blue values were calculated. Normalization calcu-
lation was achieved when each of the color was divided into
the total value [32].
After the noise reduction process by filtering, thresholding
was performed on the image and some regions which Feature extraction
belonged to the baby and color calibration card were seperated
from the image. In this phase, regions that were not adjacent to Colormap transformations In this step, yellow areas which
the baby and color calibration card were represented in black belonged to the baby’s important parts of the body were de-
color tone in the image. For insufficient cases to determine the tected. So, first, images were converted from RGB to YCbCr
threshold value, more specific local transformations were used and Lab color spaces [33]. In addition, with using feature
for segmentation [25]. When analyzing the color callibration extraction, the reduction process in image size was performed
card’s colors in RGB plane, vertical and horizontal edge de- [34]. At the end, the reduced image was obtained for other
tection algorithms were used [26]. stages. With using the colormap transformations, all colors of
the color calibration card and baby’s skin color were com-
Pixel similarity The regions that belonged to the baby and the pared in RGB color space.
card were obtained after segmentation, also other areas were
rearranged in black color tone. But, after segmentation, some Feature calculation In this step, the average values of the
colors lost and this occasion was clearly seen on the color images which were processed in YCbCr and Lab color spaces
calibration card. In this phase, with using pixel similarity were calculated for each color channel and at the end of the
method, missing colors were brought back on the color cali- calculation, totally 9 features were obtained from this step.
bration card [27]. During this process, when one center pixel Alongside color transformation, with using a linear color
is based on, the distance between this pixel and other neigh- gradient, the color difference value was calculated. Gradient
bors were calculated and obtained results were collected in a calculation was altered with the 3 × 3 Sobel filter and this filter
matrix. Then, the obtained matrix represented a new RGB was applied to the each channel of RGB. At the end of the
image [28]. In the new image, all missing colors were brought step, totally 3 features were obtained.
back on the color calibration card and P1 ve P2 pixels’ neigh- As a result of the feature calculation step, totally 12 features
borly relations were calculated with using the Eqs. (1) and (2), were obtained and these features were taken to be trained via
respectively. machine learning algorithms.


μðP1 ; P2 Þ ¼ exp −ð∥P1 –P2 ∥=Dn Þ2 ð1Þ Machine learning regressions

1 h i1=2
For this step, kNN (k-Nearest Neighbor) and SVR (Support
∥P1 –P2 ∥ ¼ ðR1 –R2 Þ2 þ ðG1 –G2 Þ2 þ ðB1 –B2 Þ2 ð2Þ
√3 Vector Regression) regressions were chosen to use because

Fig. 4 The process of

segmentation by using the color
calibration card [21]
J Med Syst (2016) 40: 166 Page 5 of 11 166

Fig. 5 Input images of the

system

these algorithms were successful for this system. Each feature space [38]. Then, input data was mapped to the
of the regression technique can compare the feature data space with using the potential non-linear function [39].
in a different angle. First, 2 different regression algo- For this step, totally 2 SVRs were adapted for training,
rithms were used on the dataset. Second, the output of these were kernel and non-linear sigmoid basis functions.
each regression which was based on a fixed threshold
was evaluated and bilirubin estimation process was per-
Bilirubin level estimation
formed using the threshold value and dataset. Below, kNN
(k-Nearest Neighbor) and SVR (Support Vector Regression)
In the bilirubin level estimation step, the threshold value
were explained in detail.
was chosen after multiple trials. Then, this value was cho-
sen as 1.91. For this step, first, maximum and minimum
kNN (k-Nearest Neighbor) The first algorithm was kNN
output value were based on to use. Then, when the dif-
(k-Nearest Neighbor) and for this algorithm, the coefficient
ference between these values was less than the threshold
was chosen as k = 5. This regression algorithm was quite
value, the average of the maximum and minimum values
succesfull to determine the bilirubin level locally [35].
was calculated and the output result was the system’s final
Bilirubin values and specific feature data were in this
bilirubin level estimation value [21]. Otherwise, 90 % of
regression. When a nondefined test vector was analyzed,
the difference of maximum and minimum values was cal-
k-neareast neighbors were found around the test vector in
culated and obtained result was the system’s final bilirubin
the feature dataset [36].
level estimation [21].
SVR (Support Vector Regressions) The first resgession
which was used was generally linear, but SVR algorithm
was non-linear [37]. When determining the relationships Results and discussion
between linear relationships, generally, Support Vector
Regressions were used. The aim of the regression was The first step of the system was defining input data with im-
finding a linear regression function in a high dimensional ages. Some of the input images which was loaded to the
166 Page 6 of 11 J Med Syst (2016) 40: 166

Fig. 6 Image segmentation

results

system were clearly shown in Fig. 5. In addition, in Fig. 5, maximum performance and success was obtained from the
each of the babies was numerized respectively, so analysis and system. Finally, with using segmentation, areas which were
comparisons were achieved easily. related to the baby’s body were remained and other areas of
The first stage of the system was color based balancing. the image were represented in black color tone.
The first step of the color based balancing was image segmen- The second step of the color based balancing was pixel
tation. With using segmentation, unused parts of the images similarity method. In this step, with using pixel similarity
were subtracted from the images and only important areas method, missing colors were brought back on the color cali-
were remained back. Segmentation results were represented bration card. For this step, each of the images was numerized
in Fig. 6. and obtained output images were given in Fig. 7 in detail.
When segmentation results which were given above were When the images which were given above were analyzed,
analyzed in detail, some colors lost on the color calibration with using pixel similarity method, missing colors were
card for each image. For example, when the results of baby-4 brought back on the color calibration card for each of the
and baby-22 were investigated, it was seen that segmentation images. For example, when the results of baby 4 and baby
was achieved successfully and the success rate was calculated 22 were investigated, the missing colors were brought back
as 100 %. Indeed, after segmentation process, 4 main colors successfully for these images and the success rate was calcu-
lost on the color calibration card and these areas were lated as 100 %. At the end of this process, for each of 80
rearaanged in black color tone. The reason for this occasion images, the success rate was calculated as 100 % and the
was based on the neigborly relations of the pixels in the im- process time was obtained as 1,3 s.
ages, so pixels of the missing color tones which were not The third step of the color based balancing was white
located around their neighbors were represented in black color balancing. In this step, the conditions of increasing resolution
tone in the final statement. and sharpness on the images were generally studied. Resulted
When this step was analyzed for 80 images, during the images obtained from this step were given in Fig. 8.
segmentation process, the success rate was calculated as The purpose of the white balancing was to make the image
100 % and the process time was obtained as 1.8 s and independent from ambient lightning, reflections and shadows.
J Med Syst (2016) 40: 166 Page 7 of 11 166

Fig. 7 Results of pixel similarity

method application

Fig. 8 Results of white balancing

application
166 Page 8 of 11 J Med Syst (2016) 40: 166

images were become independent from ambient lightning,

reflections and shadows.
The second stage of the system was feature extraction. In
this stage, with using colormap transformations and feature
calculation, for each of the images, the information of the
every pixel in RGB, YCrCb and Lab color spaces were trans-
ferred to the R, G, B; Y, Cr, Cb; L, a, b color channels and
then, machine learning regressions were performed on the
color channels’ dataset in MATLAB® environment.
Totally, for 9 channels and 80 images, dataset value was
calculated as 440*440*9 = 1.742.400. But, in each image,
when black areas were defined as unnecessary areas,
new dataset value was calculated as 900.000. In addi-
tion, with using a linear color gradient, color difference
on the color calibration card in RGB color space was
calculated and for R, G, B channels, this value was
obtained as 110*180*3 = 59.400. Finally, obtained fea-
ture value obtained as 959.400 was processed with
using kNN (k-nearest neighbor) and SVR (Support Vector
Regression) machine learning regressions. For this step, first,
maximum and minimum output value were based on to use.
Then, when the difference between these values was less than
the threshold value, the average of the maximum and mini-
mum values was calculated and the output result was the sys-
tem’s final bilirubin level estimation value. Otherwise, 90 %
of the difference of maximum and minimum values was cal-
culated and obtained result was the system’s final bilirubin
level estimation.
The third stage of the system was machine learning regres-
sions and kNN and SVR algorithms were performed on the
dataset. The outputs of this process were sent to the bilirubin
Fig. 9 The distribution of the system’s estimation results and general level estimation step. In this step, algorithms were used on
bilirubin results (TSB) which were obtained from standard blood test each of the feature information and totally 959.400 new fea-
around the decision boundary. (for this distribution, blue circles ture results were obtained. At the end of the machine learning
represented our system’s bilirubin estimation values, orange squares
process, maximum and minimum values were obtained in
represents general bilirubin results (TSB) which were obtained from
standard blood test); (a) The distribution of the data around the decision detail. In bilirubin level estimation step, when the difference
boundary with using SVR algorithm, (b) The distribution of the data between maximum and minimum values was based on to use,
around the decision boundary with using kNN algorithm the bilirubin level estimation decision loop was carried out. In
this step, threshold value was chosen as 1,91 after multiple
In this step, resulted images were analyzed and for example, trials. For this step, first, maximum and minimum output
when images of baby 4 and baby 22 were analyzed in detail, at value were based on to use. Then, when the difference
the end of the white balancing method, the rate of the resolu- between these values was less than the threshold value,
tion value of the pixels was increased from 92 to 100 and the average of the maximum and minimum values was
clarity and sharpness were gained to the images. When this calculated and the output result was the system’s final
step was analyzed for 80 baby images, the success rate of the bilirubin level estimation value. Otherwise, 90 % of the
white balancing was calculated as 100 % and the process time difference of maximum and minimum values was calcu-
was obtained as 2,1 s. Indeed, it was clearly seen that final lated and obtained result was the system’s final bilirubin

Table 1 F-statistical test results

Predicted Bilirubin Level Region of Interest Did RGB values normalize? Statistical F-test result

yskin, normalized Skin Yes 0.80 (p > 0.06)

yskin Skin No 0.85 (p > 0.06)
J Med Syst (2016) 40: 166 Page 9 of 11 166

results was 75 %. Finally, as a comment, maximum success

and performance were obtained from our system when kNN
algorithm was used and obtained results were highly consis-
tent with the standard blood test results.

Statistical analysis

Bilirubin level data which were obtained from neonatal jaun-

dice detection system were adapted to the F-statistical test to
evaluate the accuracy of the dataset and performance of the
system. In this step, a comparing process was performed be-
tween the system’s bilirubin level estimation results and the
standard blood test results and the similarity rate of the two
different results was clearly investigated. According to the
results, when the significance case was p > 0.06, a correlation
was found between neonatal jaundice detection system’s bili-
Fig. 10 ROC curve for neonatal jaundice detection system rubin results and standard blood test results. The compliance
rate between these values was calculated as 85 % according to
level estimation. Finally, the distribution of the estimation the F-statistical test. Obtained results were represented in
data which was around the decision boundary was given Table 1 in detail.
graphically in Fig. 9. Moreover, to determine the accuracy rate of the bilirubin
The distribution graphics were obtained specifically with levels and to calculate the system’s performance were
using SVR and kNN algorithms and when a decision bound- achieved in the other step and ROC (Receiving Operating
ary was created, bilirubin estimation results which were ob- Characteristics) analysis was used to analyze the dataset in
tained from our system non-invasively were compared to the detail. When preparing the ROC curve, measured data which
bilirubin estimation results which were obtained from the was obtained from blood test was based on to use. During this
standard blood test invasively. When specifically designed process, threshold value (ysc) was chosen as 205 μmol/l. If
graphics were analyzed in detail, estimation results which obtained value was higher than the defined threshold value, a
converged to the decision boundary were considered as true greater true positive rate (TPR) value and false positive rate
values. According to the obtained data, when linear correla- (FPR) value could be obtained. When the threshold level
tion value was equal to 0.81, for 40 jaundiced infants, bilirubin incresed, a fluctuation occured in TPR and FPR values.
estimation results obtained from our system were consistent Finally, ROC curve for this situation was shown in Fig. 10.
with bilirubin estimation results obtained from standard blood When ROC curve which was given above was analyzed,
test and the compliance rate was 85 %. Also, for easy com- threshold value was chosen as 205 μmol/l. Cut-off threshold
parisons, kNN and SVR algorithms were classified and ana- value (y) was calculated optimally with F-statistical test and
lyzed in each other respectively. With using kNN algorithm, the equation for calculation was given in (3) below.
more accurate results were obtained than SVR algorithm and J ¼ sensitivitiy þ specificity–1 ¼ TPR–FPR ð3Þ
the compliance rate between our system’s results and standart
blood test results was 85 %. For SVR algorithm, the compli- The first y point was varied between a higher TPR value
ance rate between our system’s results and standart blood test and a lower FPR value. When sensitivity value was 0.83 and

Table 2 Results of different works and neonatal jaundice detection system

Work Method Body Part Linear Correlation, r (n) Bland-Altman test: mean difference -
standart deviation (μmol/l)

Leartveravat, S., 2009 [40] Digital photography Sternum 0.86 (n = 61) N/A
Goel, M. , 2014 [21] Digital photography Sternum & Forehead 0.84 (n = 100) N/A
Vandermeer, B. 2013 [41] BiliCheck, JM-102/103 Forehead, Sternum 0.83 (n = 16), 0.83 (n = 10) 0.06 ± 29.46(n = 912), 3.80 ± 24.06
(n = 265)
Ural, B., 2015 (this study) Digital photography Skin (Normalized), 0.80 (n = 80), 0.85 (n = 40) 0.01 ± 38.61, 0.011 ± 38.40
Skin
166 Page 10 of 11 J Med Syst (2016) 40: 166

specificity value was 0.195, y value was calculated as optimally in a short time and the success rate of the system
195 μmol/l. In addition, one more specific point was analyzed was calculated as 85 %. Indeed, this value’s accuracy was
and when sensitivity value was 1 and specificity value was supported with the F-statistical test and ROC curve and anal-
0.50, y value was calculated as 160 μmol/l. Finally, 100 % ysis. Then, our system was compared with the other works
success rate and approximately 1.95 s process time of neonatal that were contain jaundice detection and different methods.
jaundice detection system were supported with ROC curve Finally, it was observed that our system had a unique and
and analysis. novel software, high quality and short processing time. If this
The final step was making comparisons and performance study is developed and adapted for the different other diseases,
analysis between our unique neonatal jaundice detection sys- for diagnosing stages, optimal and successful results can be
tem and different works that were contained different methods obtained from the system.
about jaundice. Comparison results were classified and given
in Table 2 in detail, respectively.
In Table 2, generally, the results of different works and
methods that were about jaundice and the results of our neo- References
natal jaundice detection system were listed in a table format.
First of all, in 2009, S. Leartveravat and friends [40] achieved 1. Click, R., Dahl-Smith, J., Fowler, L., DuBose, J., Deneau-Saxton,
to detect jaundice with 61 subjects and using digital photog- M., and Herbert, J., An osteopathic approach to reduction of
readmissions for neonatal jaundice. Osteopathic Family Physician.
raphy method for sternum. According to the results, a corre-
5(1):17–23, 2013.
lation was found between obtained results and this value was 2. Brown A.K., Kernicterus: Past, Present, and Future. NeoReviews.
calculated as 0.86. Secondly, M. Goel and friends [21] Am. Acad. Pediatrics. 4 (2), 2003.
achieved to detect jaundice with 100 participants and using 3. Madlon-Kay, D. J., Recognition of the presence and severity of
digital photography method for sternum and forehead and at newborn jaundice by parents, nurses, physicians, and icterometer.
Pediatrics. 100 (3), 1997.
the end, a correlation was found between obtained results and
4. Garfunkel, L.C., Kaczorowski, J., Christy, C., Mosby's pediatric
this value was calculated as 0.84. Thirdly, B. Vandermeer and clinical advisor: instant diagnosis and treatment. Elsevier Health
friends [41] achieved to detect jaundice with a limited number Sciences. 2002.
of participants and using measurement methods for forehead 5. Arias, I.M., Gartner, L.M., Seifter, S., and Furman, M., Prolonged
and sternum and at the end, a correlation was found between neonatal unconjugated hyperbilirubinemia associated with breast
feeding and a steroid, pregnane-3 (alpha), 20 (beta)-diol in maternal
obtained results and this value was calculated as 0.83 for each milk that inhibits glucuronide formation in vitro. J. Clin. Invest.
of them. Fourthly, in our neonatal jaundice detection system, 43(11):2037–2047, 1964.
we achieved to detect neonatal jaundice in the optimal time for 6. Murphy, J.F., Hughes, I., Verrier Jones, E.R., Gaskell, S., and Pike,
totally 80 test subjects and with using digital photography A.W., Pregnanediols and breast-milk jaundice. Arch. Dis. Child. 56:
474–476, 1981.
method babies’ skin and the color differences were analyzed.
7. Hafkamp, A.M., Oral treatment of unconjugated hyperbilirubinemia.
At the end of the process, a correlation was found between Ponsen & Looijen B.V, Wageningen, The Netherlands, 2006.
obtained results and this value was calculated as 0.85. This 8. McDonagh, A.F., Movement of bilirubin and bilirubin conjugates
system has advanced methods and techniques and also, ac- across the placenta. Pediatrics. 119(5):1032–1033, 2007.
cording to the obtained results, neonatal jaundice was detected 9. Kramer, L.I., Advancement of dermal icterus in the jaundiced new-
born. Amer. J. Dis. Child. 118:454–458, 1969.
optimally with a high performance.
10. Burke, B., Robbins, J., and Hobbs, C., American Academy of
Pediatrics Subcommittee on hyperbilirubinemia. Management of
hyperbilirubinemia in the newborn infant 35 or more weeks of
Results gestation. Pediatrics. 114(1):297–316, 2004.
11. Cremer, R.J., Perryman, P.W., and Richards, D.H., Influence of
light on the hyperbilirubinaemia of infants. The Lancet, Elsevier.
In this study, the aim was creating a non-invasive system to 271(7030):1094–1097, 1958.
detect and control the jaundice and helping doctors for early 12. Cappellini, M.D., Di Montemuros, F.M., Sampietro, M., Tavazzi,
diagnosis. Generally, when 40 healthy infants were based on D., and Fiorelli, G., The interaction between Gilbert's syndrome and
to use, for 40 jaundiced neonatals, the results which were G6PD deficiency influences bilirubin levels. Br. J. Haematol.
104(4):928–929, 1999.
obtained from the flowchart were compared to the standart
13. Newman, T.B., Escobar, G.J., Gonzales, V.M., Armstrong, M.A.,
blood test measurements. Because of the sensitivity of the Gardner, M.N., Folck, B.F. Frequency of neonatal bilirubin testing
newborn babies and the courtesy of the vessels, our proposed and hyperbilirubinemia in a large health maintenance organization.
neonatal jaundice detection system was used commonly to Pediatrics. 104 (6), 1999.
detect the jaundice non-invasively on time and this system 14. Stokowski, L.A., Fundamentals of phototherapy for neonatal jaun-
dice. Adv. Neonatal. Care. 6(6):303–312, 2006.
gave users accessibility and short processing time for detec- 15. Geifman-Holtzman, O., Wojtowycz, M., Kosmas, E., and Artal, R.,
tion. When the processes of the system were investigated, for Female Allo-immunization with antibodies known to cause hemo-
40 jaundiced babies, jaundice was detected successfully and lytic disease. Obstet. Gynecol. 89(2):272–275, 1997.
J Med Syst (2016) 40: 166 Page 11 of 11 166

16. Mollison, P.L., Engelfriet, C.P., and Contreras, M., Blood transfu- 30. Yule, J.A.C., Principles of color reproduction. Wiley, New York,
sion in clinical medicine, 10th edn. Blackwell Science, Oxford, 1967.
1997. 31. Viggiano, J.A.S., Comparison of the accuracy of different white
17. Shapiro, L.G., and Stockman, G.C., Computer vision. Prentice- balancing options as quantified by their color constancy. Sensors
Hall, New Jersey, pp. 279–325, 2001. and camera Systems for Scientific, industrial, and digital photogra-
18. Rosenfeld, A., and Kak, A.C., Digital picture processing. Academic phy applications V: proceedings of the SPIE. Bellingham. WA:
Press, San Diego, 1982. SPIE: the International Society for Optical Engineering. 5301:
19. Hsien-Che, L., Introduction to color imaging science. Cambridge 323–333, 2004.
University Press. pp 450, 2005. 32. Petschnigg, G., Szeliski, R., Agrawala, M., Cohen, M., Hoppe, H.,
20. Sharma, G., and Bala, R., Digital color imaging handbook. CRC and Toyama, K., Digital photography with flash and no-flash image
Press, Boca Raton, 2002. pairs. ACM Trans. Graph. 23(3):664–672, 2004.
21. De Greef, L., Goel, M., Seo, M.J., Larson, E.C., Stout, J.W., Taylor, 33. Schwarz, M.W., Cowan, W.B., and Beatty, J.C., An experimental
J.A., Patel, S.N., Bilicam: using mobile phones to monitor newborn comparison of RGB, YIQ, LAB, HSV, and opponent color models.
jaundice. UbiComp '14 Proceedings of the 2014 ACM International ACM Trans. Graph. 6(2):123–158, 1987.
Joint Conference on Pervasive and Ubiquitous Computing. 331– 34. Florack, L., and Kuijper, A., The topological structure of scale-
342, 2014. space images. J. Math. Imaging Vision. 12(1):65–79, 2000.
35. Liu, D.Y., Chen, H.L., Yang, B., Lv, X.E., Li, L.N., and Liu, J.,
22. Forcade, N., Le Guyader, C., and Gout, C., Generalized fast
Design of an Enhanced Fuzzy k-nearest neighbor classifier based
marching method: applications to image segmentation. Numerical
computer aided diagnostic system for thyroid disease. J. Med. Syst.
Algorithms. 48(1):189–211, 2008.
36(5):3243–3254, 2012.
23. Horowitz, S.L., Pavlidis, T., Picture segmentation by a directed split
36. Coomans, D., and Massart, D.L., Alternative k-nearest neighbour
and merge procedure. Proc. ICPR. Denmark. 424–433, 1974.
rules in supervised pattern recognition: Part 1. k-Nearest neighbour
24. Haddad, R.A., and Akansu, A.N., A class of fast gaussian binomial classification by using alternative voting rules. Anal. Chim. Acta.
filters for speech and image processing. IEEE Trans. Acoust. 136:15–27, 1982.
Speech Signal Process. 39(3):723–727, 1991. 37. Nouira, K., and Trabelsi, A., Intelligent monitoring system for in-
25. Haindl, M., Mikes, S., Texture segmentation benchmark. Proc. of tensive care units. J. Med. Syst. 36(4):2309–2318, 2012.
the 19th Int. conference on pattern recognition. IEEE Computer 38. Berikol, G.B., Yildiz, O., and Özcan, İ.T., Diagnosis of acute coro-
Society. 1–4, 2008. nary syndrome with a support vector machine. J. Med. Syst. 40:84,
26. Chan, T.F., and Vese, L., Active contours without edges. IEEE 2016.
Trans. Image Process. 10(2):266–277, 2001. 39. Cortes, C., and Vapnik, V., Support-vector networks. Mach. Learn.
27. Pratt, W.K., Digital image processing, 4th edn. Hoboken, John 20(3):273, 1995.
Willey & Sons, 2007. 40. Leartveravat, S., Transcutaneous bilirubin measurement in full term
28. Demirci, R., Rule-based automatic segmentation of color images. neonate by digital camera. Medical Journal of Srisaket Surin
AEU Int. J. Electron. Commun. 60(6):435–442, 2006. Buriram Hospitals. 24(1):105–118, 2009.
29. Funt, B., Cardei, V., Barnard, K., Learning color constancy. 41. Nagar, G., Vandermeer, B., Campbell, S., and Kumar, M., Reliability
Proceedings of the Fourth IS&T/SID Color Imaging Conference, of transcutaneous bilirubin devices in preterm infants: a systematic
58–60, 1996. review. Pediatrics. 132(5):871–881, 2013.
Reproduced with permission of copyright owner. Further reproduction
prohibited without permission.