Introduction Motivation For The Project

Published Date: 02 Nov 2017

Chapter 1

A spirometer is a tool that can be used to measure respiratory volume and flowrate. Typical spirogram plots the expiratory air flow against the total expiratory volume. This information is commonly used to diagnose chronic obstructive pulmonary disease, or COPD.

The main motivation of this project is to develop an incentive based sensor for pediatric population as there are very few incentive based spirometers for the age group below 15 years and hence spirometry is not a reliable test in this population. Presence of an incentive based sensor will help in an infants to perform the FEV1 maneuver better. where it gives the air volumes for first second based on human exhalation and it is child’s friendly, making it more suitable for smaller children to use.

Another factor influencing the efficiency of COPD treatment is the high potential

for variability between tests. Traditionally, a patient performs the spirometry maneuvers

while being monitored and instructed by a trained technician. Unfortunately, the quality

of coaching provided by different technicians can lead to significantly different results.These variances have the potential to be even more significant if the patient is monitored at various facilities.

.

1.2 Problem Definition

This study would be to step towards detection of pulmonary disorders in an easier way and to make children to blow comfortably. The sensor which we are going to design is cost effective, easy to operate, small and portable. It is a child’s friendly since we are developing an incentive based sensor which attracts the children to blow and cooperate with the instructor.

1.3 Objective

This work aims to describe a bone age assessment method based on the lengths and widths of the metacarpals and Metacarpal index. We also aim to reduce the human interference in the age assessment process using this method.

an incentive based sensor development for performing FEV1 maneuver in children based on human exhalation and we also aim to reduce the size of the mouthpiece to make it more suitable for smaller children to use.

Chapter 2

Literature survey

2.1. Anatomy of Hand

2.1.1 Skeleton

The human hand skeleton consists of 27 bones. These include 8 carpals, 5 metacarpals and 14 phalangeal bones.

http://visual.merriam-webster.com/images/human-being/anatomy/skeleton/hand.jpg

The 8 carpal bones are scaphoid, lunate, triquetral, pisiform, trapezium, trapezoid, capitate and hamate. These carpals in turn join with the two forearm bones, i.e, radius and ulna to form the wrist joint.

The carpals bones articulate with the bases of metacarpal bones to form the palm. The metacarpal bones are five in number and these connect to each finger and thumb. These bones are then connected to phalanges to form the knuckle or metacarpophalangeal joints (MCP joints).

There are three phalanges in each finger except the thumb. The one that is closest to the knuckle joint is called the proximal phalanx. The bone at the end of the finger is called the distal phalanx and the middle bone is known as intermediate phalanx. Hence in total there are 14 phalangeal bones.

2.1.2 Ossification

Ossification is the natural process of bone formation. There are two kinds of ossification namely endochondreal ossification and intra-membranous ossification. Most of the bones are formed by endochondreal ossification where ‘endo’ means within and ‘chondreal’ means cartilage. In this method bones are formed from hyaline cartilage as seen in stage one. In the second stage, periosteum leads to the creation of a bone collar which combines on to the shaft portion and chondrocytes at the centre of the cartilage signals the surrounding matrix to ossify. The cartilage matrix breaks down and cavity is formed. Hence this region is called the primary ossification centre. The third step is the invasion of cavity or the periosteal bud. In this step we observe that the periosteal bud located at the centre of cartilage is made of arteries (which carry nutrients), veins, cells (for formation of bone marrow & osteogenic stem cells) and osteoblasts for bone formation. Spongy bones are formed around the remaining cartilage matrix. The secondary ossification centre appears at each end of the epiphysis in this step. In the fourth stage, elongation of epiphysis or shaft occurs. In this step cartilage matrix ossification occurs, cartilage cells die, periosteal bud invasion occurs and spongy bones are formed. In the fifth stage further ossification & remodelling occurs. In this stage further maturation of bones occurs and cartilage is eliminated. Only epiphyseal plate and articular cartilage are retained.

C:\Documents and Settings\MARTINA\My Documents\Downloads\download.jpg

The bone consists of the following parts:

Epiphysis

Diaphysis

Metaphysis

Epiphysis: The ends and the tips of a long bone that ossify from secondary centers are known as epiphyses.

Diaphysis: Diaphysis is the term used for elongated shaft of the long bone. It ossifies from a primary center.

Metaphysis: The ends of diaphysis near the epiphyses are known as metaphysis.

When the child grows, the epiphysis starts getting fused to the diaphysis. This continues till the age of 18 or 21.

2.2 Bone Age Assessment

Bone age assessment (BAA) is a non-invasive clinical procedure used to evaluate the stage of skeletal maturation based on a left hand radiograph of the subject.

2.2.1 Definition of bone age

A person’s age measured by matching their bone development (as shown by X-rays) with the bone development of a normal person of known chronological age). A difference between chronological age and skeletal age may suggest abnormalities in skeletal development.

2.2.2 Existing Clinical Methods

2.2.2.1 Greulich-Pyle method

The Greulich and Pyle method (GPmethod) is an atlas-basedmethod. The assessment is performed by comparing a person’s left hand-wrist radiograph with an atlas containing standard radiographs for a range of skeletal ages in which the development stages of each bone is described in 2 different atlas separately for males & females. The method requires the assessor to work systematically through the chosen radiograph, comparing each of 31 bones and sesamoids in the hand-wrist with the standards in the atlas that varies with the race, age and sex.Then an age is assigned to each bone using data from tables associated with the standard that contains the closest match to the bone. If no match is found for a bone, the age is estimated from the closest matching radiographs. Each radiograph also has a written description of important skeletal maturity indicators for the associated skeletal age. If the age of each bone corresponds to a single standard in the atlas then the person’s bone age is that of the standard. If there is no exact match and the results are intermediate between two standard radiographs, then an intermediate estimate between the two corresponding ages is used.

Figure: A radiologist reporting a bone age radiograph using the atlas-based Greulich and Pyle method

2.2.2.2 Tanner-Whitehouse method

The Tanner-Whitehouse method is also known as bone ossification test. It is based on left hand radiographs. In this method twenty regions of interest namely distal radius, distal ulna, first, third and fifth metacarpals, proximal phalanges of the thumb, third and fifth fingers, middle phalanges of the third and fifth fingers, distal phalanges of the thumb, third and fifth fingers, the seventh carpal bones: capitate, hamate, triquetral, lunate, scaphoid, trapezium and trapezoid are used. Fig.1 shows some of the bones of interest. This method uses a detailed analysis of each individual bone. Each ROI is divided in three parts namely epiphysis, metaphysis and diaphysis as shown in fig.1 (b).

The development of each ROI is represented by letters A, B, C, D, E, F, G, H and I. Stage A represents bone is absent, Stage B represents single deposit of calcium, Stage C represents center is distinct in appearance, Stage D represents maximum diameter is half or more the width of metaphysic, Stage E represents border of the epiphysis is concave, Stage F represents epiphysis is as wide as metaphysis, Stage G represents epiphysis caps the metaphysis, Stage H represents fusion of epiphysis and metaphysis has begun and Stage I represents epiphyseal fusion completed as shown in fig.2.

Then a numerical score is associated with each stage of each bone as shown in table 1. By adding the scores of all ROIs, an overall maturity score is obtained. This score is correlated with the bone age differently for males and females.

TW2 was a revision of TW1, especially in relation to the scores associated to each stage and also the difference between both sexes. Then TW3 method came into existence which utilized thirteen RUS (including radius, ulna and short finger bones) ROIs and seven carpal ROIs.

2.2.3 Applications

The most common applications of bone age assessment are in the diagnosis of growth disorders ,prediction of adult height,monitoring growth hormone treatment,in forensics and sports.

2.2.2.1 Forensics and Sports

Bone age is used by forensic scientists to confirm one’s age,especially in case of criminals & when the date of birth of the child is unknown or cannot be confirmed.It can also be used in sports to estimate the age of athletes.

2.2.3.2 Diagnosis of growth disorders

There are various causes for abnormal growth in children.They can be classified as hormonal and non-hormonal causes.Most of these causes can be diagnosed through clinical symptoms and blood test.But the major exception in this case is growth hormone deficiency.This is very difficult to diagnose using blood test because of pulsatile secretion of growth hormone into blood stream.This has a great impact on growth and development of skeleton.So usually in diagnosing growth disorders,the clinical measurements like height & width is done first.By combining these with the child’s bone age ,it is possible to differentiate among the growth disorders.If there is any defect in skeletal syatem,there is slow bone growth.

2.2.3.3 Prediction of adult height

Bone age can be used to predict the final adult height.This is very important in the managements of growth disorders & psychosocial well-being of the child.

2.2.3.4 Monitoring growth hormone treatment

Growth failure & growth hormone deficiency are treated with synthetic human growth hormone.But this is expensive in most of the countries.Therefore,its necessary to give as much as growth hormone required.Bine age is used by most of the physicians in deciding when to start & stop treatment.Bone age observations help in monitoring the response to growth hormone.

Chapter 3

Metacarpal Index Method

3.1 Introduction

In this work, we use Metacarpal Index i.e MCI to estimate the age of a person.MCI can be calculated using any of the below mentioned definitions

Average of the length and width ratio of 2-5 metacarpals.

Ratio of sum of lengths by sum of widths of 2-5 metacarpals.

MCI=(A+B+C+D)/(a+b+c+d)

Here ,we create a database which consists of MCI and other features such as length and width of individual metacarpals. This is done by implementing various image processing techniques on left-hand radiographs, followed by respective measurements. This kind of database is currently available for African and American origin. But then there is no such database for Indian origin. Hence, in this work we create separate databases for both males and females.

Later using this database, given a left-hand radiograph of unknown age,the MCI and other features of metacarpals are evaluated. Then with the help of neural network classifier, the age of a person can be estimated.

3.2 Comparison with existing clinical Methods

The GP Atlas method involves comparing the whole image with an atlas, while the TW method uses scoring of each bone. But in MCI we take only the region of Interest, i.e., the metacarpals and preprocess it to determine the age.

The GP method uses 28 growth points, TW uses 13 growth points but MCI uses only 4 bones. GP atlas is prone to human error because it is a manual method but MCI has less error rate compared to its predecessors. The Tanner Whitehouse method can be used only up to the age of 18-21 but MCI can be used above these age groups.

Chapter 4

Methodology

Image acquisition

The hand-wrist radiograph is formed from x-rays leaving an x-ray source, passing through the hand, and some of the transmitted x-rays causing ionisation in an x-ray sensitive receptor. The x-rays usually pass through the hand in a straight line. Of those that do not, some undergo scattering, and the rest are completely absorbed by bone or soft tissue. The amount of x-ray scattering and absorption depend on the energy of the x-rays and the thickness, density, and atomic number of the material through which they pass.

The type of x-ray image receptor can also influence the contrast. To expand the optimal-contrast range, traditional screen-film combinations can be replaced by computed radiography or direct radiography. Computed radiography uses photostimulable phosphor plates that store the ionisation caused by the x-rays as a latent image, that is then readout from the plate using a laser beam to stimulate the phosphor. Alternatively, a direct radiography panel can be used to immediately convert the x-ray ionisation to an electrical signal for storage as digital image.

For computerised bone age assessment - the image exposed upto 0.5mW is acquired in AP-position(in which the beams pass from front-to-back) and stored in digital form i.e. ready for transfer to a processing system. The digital form of the radiographic image is usually a rectangular array of pixels with depths of 8, 12 or 16 bits.

The overall contrast in the image is determined by the energy of the x-rays, scattered and absorbed radiation in the tissues of the hand, energy-dependent absorption of x-rays at the image receptor, and the relationship between the x-ray flux reaching the receptor and the output pixel values.

The overall image resolution is determined by the size of the x-ray source in the x-ray tube, the distance from the x-ray tube to the image receptor, the distance between the hand and the image receptor, and the design of the image receptor itself.

The noise in the digital image is determined by a combination of structural noise sources from the image receptor, variations in the x-ray fluxes due to stochastic processes of x-ray generation and interaction (quantum mottle), and quantization noise from the digitization process.

Overview of the process

This is the overview of the process. The image is acquired in DICOM format. DICOM (Digital Imaging and Communications in Medicine) is a standard for handling medical images. DICOM files contain metadata that provide information about the image data, such as the size, dimensions, bit depth, modality used to create the data, and equipment settings used to capture the image. Here, we use the same image as the input so that the original features of the image are preserved.

Pre-processing

Pre-processing is the process of enhancing images prior to further processing. It includes background noise removal, normalizing the intensity values, masking the portions of image etc. The pre-processing requirements depend on the image. In this stage, pre-processing includes normalization and selecting regions of interest.

Normalization is done to scale down the intensity values between 0 and 1 that is described as follows,

Normalization=

The region of interest (ROI) i.e. metacarpals (2-5) is cropped.

Thresholding

Thresholding is a non-linear technique that converts a gray-scale image into a binary image where the two levels are assigned to pixels that are below or above the specified threshold value. It can be illustrated as follows,

If a [m, n]>=T a [m, n] =1

Else a [m, n] =0

where T is the maximum threshold value

a [m, n] is the image.

In this step, since the intensity of soft tissue is lower than bones, thresholding is used to eliminate soft tissue and retain only bones. Otsu’s thresholding is preferred over other thresholding techniques because it is a type of global based thresholding that automatically performs histogram shape-based image thresholding, or, the reduction of a gray-level image to a binary image. The algorithm assumes that the image to be thresholded contains two classes of pixels or bi-modal histogram (e.g. foreground and background) then calculates the optimum threshold separating those two classes so that their combined spread (intra-class variance) is minimal. This type of thresholding works on the normalized value. The algorithm is as follows,

Consider the normalized histogram as a probability distribution function as in,

Pr (rq)=nq / n q=0,1,2, . . . . (L-1)

Where n is the total number of pixels in an image

nq is the number of pixels that have intensity level rq.

L is the total number of possible intensity levels in an image.

Suppose a threshold k is chosen such that

C0 is the set of pixels with levels 0, 1, . . . . , k-1

C1 is the set of pixels with levels k, k+1, . . . ., L-1

Otsu’s thresholding method choose a thresholding value k that maximizes the inter class variance which is defined as,

where

After thresholding the intensity value of the bones which are similar to that of soft tissues are eliminated which results in lot of unwanted pixels in the is negated to fill the present.

Segmentation of metacarpals

Segmentation of an image is a process of sub-dividing the image into its constituent objects or regions. The level of segmentation depends on the problem. Segmentation should stop when the objects of interest are isolated. In this stage, metacarpals are segmented from other bones for further processing.

Setting a baseline

The baseline is set above the point where the metacarpals are fused with the carpals. The image is cropped till the baseline to separate the metacarpals into four different connected components. This eliminates the distal end of the metacarpals but the length of the metacarpal remains unchanged in further steps. The value of baseline is user dependent which in turn depends on the image.

Segmentation using area based statistics

In this step, the number of connected components is found. The area of all the components are computed .The first four maximum areas are considered to be the areas of metacarpals 2-5,these are than set to zero by an iterative process .The XOR operation of this image and the binarized image ,segments the metacarpals from the phalanges.

Bounding box and object orientation

The boundaries of the segmented metacarpals are extracted in this stage. Then a bounding box is masked onto the cropped image to segment each metacarpal automatically. A bounding box is a 2-by-2 double array that specifies the minimum (row 1) and maximum (row 2) values for each dimension of the image data. This process is repeated for all four metacarpals. In this way we are able to segment each metacarpal.

The segmented image is then rotated to an angle of 90° for reducing slope inaccuracy and getting a proper line segment for estimating the length and width. Orientation is a scalar and it denotes the angle (in degrees ranging from -90 to 90 degrees) between the x-axis and the major axis of the ellipse that has the same second-moments as the region. This property is supported only for 2-D input label matrices.

Measurements

Width

After aligning the metacarpals, the shaft of the four metacarpals is cropped by selecting the two extreme points of the shaft. The transpose of the shaft is taken and the sum of ones along each column is computed. The minimum of the sum is considered to be the width of the shaft. This is then mapped on to the aligned metacarpal, the width is located and measured in terms of pixels. This is later converted to millimeters using the formula

width_mm=w/(300*25.4)

where ,w=width of metacarpal in pixels

4.2.4.2 Length

Transpose of the aligned metacarpal is used to estimate the length. Here the sum of one’s of transposed aligned metacarpal is computed along each column. The first non-zeroth column is assumed to be the tip of the metacarpal. The distance from the midpoint of the width to the tip of metacarpal is found. Twice of this distance is the length of the metacarpal. This is in terms of pixels. This is again converted to millimeters using the formula

length_mm=l/(300*25.4)

where ,l=length of metacarpal in pixels

4.2.4.3 MCI

The length and width of each metacarpal is considered for MCI measurement. Metacarpal index can be measured using any of the earlier mentioned formulae.In this work, MCI is considered to be the ratio of sum of lengths to sum of widths of 2-5 metacarpal(except thumb).This can be given as follows:

MCI= sum of the lengths (2-5)

Sum of the widths (2-5)

4.2.5 Database creation

4.2.6 k-Nearest Neighbour or kNN classifier

This is one of the simplest classifier.It is an Instance based classifier.The ‘k’ in kNN classifier refers to the fact that the algorithm looks for k different neighbours during classification.By default the value of k is 1.In this classifier the mean value for each training data is calculated.This process is known as learning.Later if a new data comes along whose input value(mean) is known,but we wish to classify.Then,first the k nearest neighbours should be found.And the majority class label is returned.The value of k in this chapter is 1.So basically the Euclidean distance between the testing data and all the trained data is calculated.The class which corresponds to smallest distance is returned as the class label i.e the nearest neighbor in instance space.

Chapter 5

Results

5.2 Results

Comparing the FEV1 values of an Incentive based Sensor with the Schiller Spirometer.

FEV1 Values obtained from Schiller Spirometer

FEV1 Values obtained from Incentive Based Sensor

2.71

3.5

2.74

2.4

2.72

3.6

2.62

3.3

2.83

3.2

1.9

2.8

Table 5.2: Comparing FEV1values of an Incentive based sensor with the Schiller Spirometer

Description

As shown in the above table 5.2, the FEV1 values which we obtained from an Incentive based Sensor is not exactly matching with the Schiller Spirometer. We can see there is little deviations in the FEV1 values and this is because of some limitations of the Hall effect Sensor and other factors are also considered, which is explained in chapter 6.

Chapter 6

Conclusion

6.1 Conclusion of Bone Age Assessment

This work aims to develop an incentive based sensor for pediatrics. As there are many spirometers available for the adults but very few incentive based spirometers are there for pediatrics population and hence spirometry is not a reliable test in this population. Therefore, the presence of an incentive based sensor will help the child to do FEV1 maneuver in an easier way.

The progress we have made this semester is developing an incentive based sensor which will measures the air volumes for first second and last second. From table 5.2 , we can see there is little deviations in the FEV1 values of an incentive based sensor , which is compared with the ‘GOLD STANDARDS’ . This deviations are due to the following reasons:

The shape of the fan

The type of the materials used

The diameter of the pipe through which the air passes

Due to the limitations of Hall effect sensor

The magnet is attached to the fan wing, so that the fan can rotate upto 2-3 seconds but not upto 6 seconds because of the magnet weight.

It also depends on the magnetic properties.

Current Spirometers will give FEV1/FVC(for 6 sec) but Incentive based sensor will give FEV1/FVC( for 2-3 sec), where FVC values for 6 second cannot be measured using an Incentive based sensor, until and unless if a person blows for 6 seconds. Due to the magnet weight, fan will stops rotating after 2nd or 3rd second, so IBS cannot measure up to 6th second. Therefore the incentive based sensor (IBS) which we have developed has to be upgraded. By using the proper materials, shape of the fan and other factors which are responsible for the limitations has to be altered, in order to meet its Gold Standards.

Therefore this project can be a reward for the pediatrics, inorder to perform FEV1 maneuver. Incentive based sensor (IBS) also consists of an mobile fan toy, which will attract the child to blow very effectively to get FEV1 values. Since it is a prototype but practically cannot be used on infants because the device has to be upgraded inorder to meet the ‘Gold Standards’.

FUTURE WORK

Suitable materials can be used to upgrade the device.

The fan shapes can be altered, in order to rotate the fan for about 6 seconds.

The magnet size can be reduced or any other alternatives can be used.

References

[1]  "Broadribb's Introductory Pediatric Nursing". Nancy T. Hatfield (2007). p.4. ISBN 0781777062

[2] Pearson, Howard A. (1991). "'Pediatrics in the United States'". In Nichols, Burford L. et al.(eds). History of Paediatrics 1850–1950. Nestlé Nutrition Workshop Series. 22. New York, NY: Raven Press. pp. 55–63. ISBN 0-88167-695-0.

[3]  "Children are not just small adults: the urgent need for high-quality trial evidence in children". PLoS Medicine. Retrieved 2009-09-04

[4] Notes on human respiratory system physiology (Dr. Gulerdemli)

[5] http://kidshealth.org/parent/system/medical/spirometry.html#

[6] http://www.nexegen.net/nexegen-electronics/sensors-HALL-EFFECT-SENSOR.php

[7] M. A. Mazidi and J. G. Mazidi, The 8051 Microcontroller and Embedded System. Pearson Education, India, 2003.

[8] Kenneth J.Ayala, The 8051 Microcontroller Architecture programming and Application. Penram International Pub; Second Edition: Mumbai, 2003.

[9] http://www.engineersgarage.com/electronic-components/max232-datasheet

[10] http://www.wisegeek.com/what-is-a-spirometer.htm