3d e-lab | Personal Website of Glen Turley
Hi all, it has been a while since my last blog post. I was busy lecturing and working away, then I was enjoying myself in Vietnam and then I was lazy. I did read some excellent books during this time. Two of which particularly stood out, the new book by Chris Anderson entitled "Makers: The New Industrial Revolution" and "Skunk Works" by Ben Rich. They were written some years apart, but in a way they are related. Skunk Works is an excellent tale about how innovation can thrive within a big company to achieve some marvellous feats of engineering, while "Makers" brings us up to date with how design and innovation is being democratised through the new generation of manufacturing tools, such as 3d printer, laser cutters, and desktop CNC machines. It also provides insight with regard to open hardware business models and financing so you can now become a manufacturer start-up.
Away from my reading, I have spent some time this bank holiday weekend trying to solve the mystery of how to read the sensor data from both the MPU6050 and HMC5883L. Before I went away I had successfully got some code up and running to display the raw accelerometer and gyro values of the MPU6050. However, I was tying myself in knots with how to read the heading data from the HMC5883L. Needless to say things got far too complicated and the whole thing was looking a mess. This weekend, I started from scratch and the whole endeavour was far less taxing. I integrated the HMC5883L libraries developed by Jeff Rowberg with the MPU6050 c++ files and then enabled pass-through mode on the MPU6050 so that the HMC5883L could be accessed. The result was a successful connection with all 9 degrees of freedom information now being displaying. My next task, is to understand the sensor calibration and then I think I can start doing things with the sensor! The hacked code is available [here]
A couple of weeks ago, I showed you my first attempt at assembling and soldering my IMU device. Unfortunately, my attempt to reflow was not that successful and I ending up cooking one of the chips. You see across from you to the left, my third and fourth attempts – these were a lot more successful! I managed to hook them up to my Raspberry Pi and check the I2C communication signal and. . . . my Pi and my IMUs recognise each other. I am now looking forward to doing more with the IMU and to see whether I can start getting the raw accelerometer, gyro and magnetometer readings printing out on my Pi. One person who is already doing as such and even has a superb graphical display functioning is David Grayson [link]. I suggest you check out his site to see what else he is up too.
There have been one or two other things that I have been up to, the first is producing another medical model shown below. I have been using Simpleware to get some fantastic results and the meshes produced are really clean and easy to work with. As you can see, I managed to edit the below model before 3D printing to connect the patella (knee cap) and the proximal and distal ends of the femur (thigh) bone. I have also been trying to simplify a mesh for the Formula Student team at my University with limited success at present because of the size of complexity of the dataset. While I was looking for solutions, I was side tracked looking at 3d printers and laser scanners and the quality you get for the price is really coming towards the ‘prosumer’ segment. Check out the CubeX (derived from a Bits from Bytes 3D Touch) [link] and the NextEngine [link] and see what I mean. It is at the stage where you can commence your own digital prototyping business for a reasonable cost, in short the equipment is become affordable to the engineers, scientists and designers who use them the most and therefore derive the most benefit. This can only be good and I see exciting things happening.
I had a mixture of good and bad news with regard to my research over the christmas break. The bad news was that the fellowship application that I submitted was not accepted. However, I got some very positive feedback and will now concentrate on putting together a new and improved fellowship application. The good news is that I had a journal paper accepted and published online in Medical & Biological Engineering & Computing. The whole process from submission to acceptance took in total one month, which is the fastest that I have experienced. The journal paper is available at the following [link] and uses a CT simulation to evaluate range of motion restriction within the hip joint. The topic has been the subject of a previous blog post available [here]
Just before christmas all of the components for my IMU device arrived. Last week, I finally made an attempt at assembling the circuit after having a very relaxing break. Unfortunately, I was a little too aggressive with the application of the solder paste and managed to distort the solder stencil. Today, I had a second attempt with a new stencil and managed to successfully apply the paste onto the pcb and assemble the components. I don't have a reflow oven, so I used a homemade solution which I am not sure worked that well. I will know once I have soldered the jumper and have done some testing. I have more pcbs and components to be able to assemble another device, but next time I will see whether I can use a reflow oven. I am confident that I will have a working device very shortly. The learning curve is not that steep and I would encourage you all to have a go. There is so much help and guidance online, I am sure you will be successful whatever your target is. Unfortunately, there will be one person who can no longer help us. Fabio Varesano of www.varesano.net passed away over Christmas at such a young age. His work on developing an IMU device was way ahead of mine and he made his work very accessible and engaging.
I had a very strange request last week from one of the clinical research fellows I collaborate with. They asked whether I could come in to help develop a protocol for MR imaging of the hip to assess the shape of the pelvic acetabulum and femoral head. If the shape of either is abnormal then this can cause a patient pain and limit their ability to perform daily activities, particularly if they are very active. I initially thought that this was a perfectly normal request, until I found out that I was the person going into the MRI! It was actually quite a relaxing experience lying there as my hip was being scanned early on a Friday morning. I then received the scan data to analyse, which felt strange considering it was my own hip. I used Simpleware ScanIP to segment the MR Scan and reconstruct the slices into a 3D Nurbs Model. You can see the results in the picture across from this text. I was pleased with the results considering that it was a 2mm spacing between the slices. We are also looking at a fast analysis route where we can build a picture of the three-dimensional hip shape abnormality directly from the 2D slices. To do this, I used ImageJ to reconstruct the images in an oblique plane, perpendicular to the femoral neck axis. I could then slice through this image stack at regular increments around the centre of the femoral neck axis. The results are shown in the images below, where using a clock face I have resliced the images in hourly increments to produce images which provide an overview of the circumferential contour of the femoral head-neck junction. These can be used to measure individual alpha angles, to assess the asphericity of the femoral head. Usually, an alpha angle of greater than 55 degrees signifies to the surgeon that the hip is abnormal. However, in a journal article I am currently writing with orthers, we shows that a normal hip can also have this degree of asphericity. To be able to diagnose an abnormal hip the entire circumference of the femoral head-neck junction must be considered in conjunction with an assessment of the daily activities one undertakes. Once I have developed a reliable process to reconstruct these radial images I will post a video showing how it can be done in ImageJ.