Methodological improvement of 3D fluoroscopic analysis for the robust quantification of 3D kinematics of human joints

3D video-fluoroscopy is an accurate but cumbersome technique to estimate natural or prosthetic human joint kinematics. This dissertation proposes innovative methodologies to improve the 3D fluoroscopic analysis reliability and usability. Being based on direct radiographic imaging of the joint, and avoiding soft tissue artefact that limits the accuracy of skin marker based techniques, the fluoroscopic analysis has a potential accuracy of the order of mm/deg or better. It can provide fundamental informations for clinical and methodological applications, but, notwithstanding the number of methodological protocols proposed in the literature, time consuming user interaction is exploited to obtain consistent results. The user-dependency prevented a reliable quantification of the actual accuracy and precision of the methods, and, consequently, slowed down the translation to the clinical practice. The objective of the present work was to speed up this process introducing methodological improvements in the analysis. In the thesis, the fluoroscopic analysis was characterized in depth, in order to evaluate its pros and cons, and to provide reliable solutions to overcome its limitations. To this aim, an analytical approach was followed. The major sources of error were isolated with in-silico preliminary studies as: (a) geometric distortion and calibration errors, (b) 2D images and 3D models resolutions, (c) incorrect contour extraction, (d) bone model symmetries, (e) optimization algorithm limitations, (f) user errors. The effect of each criticality was quantified, and verified with an in-vivo preliminary study on the elbow joint. The dominant source of error was identified in the limited extent of the convergence domain for the local optimization algorithms, which forced the user to manually specify the starting pose for the estimating process. To solve this problem, two different approaches were followed: to increase the optimal pose convergence basin, the local approach used sequential alignments of the 6 degrees of freedom in order of sensitivity, or a geometrical feature-based estimation of the initial conditions for the optimization; the global approach used an unsupervised memetic algorithm to optimally explore the search domain. The performances of the technique were evaluated with a series of in-silico studies and validated in-vitro with a phantom based comparison with a radiostereometric gold-standard. The accuracy of the method is joint-dependent, and for the intact knee joint, the new unsupervised algorithm guaranteed a maximum error lower than 0.5 mm for in-plane translations, 10 mm for out-of-plane translation, and of 3 deg for rotations in a mono-planar setup; and lower than 0.5 mm for translations and 1 deg for rotations in a bi-planar setups. The bi-planar setup is best suited when accurate results are needed, such as for methodological research studies. The mono-planar analysis may be enough for clinical application when the analysis time and cost may be an issue. A further reduction of the user interaction was obtained for prosthetic joints kinematics. A mixed region-growing and level-set segmentation method was proposed and halved the analysis time, delegating the computational burden to the machine. In-silico and in-vivo studies demonstrated that the reliability of the new semiautomatic method was comparable to a user defined manual gold-standard. The improved fluoroscopic analysis was finally applied to a first in-vivo methodological study on the foot kinematics. Preliminary evaluations showed that the presented methodology represents a feasible gold-standard for the validation of skin marker based foot kinematics protocols.

Flexible Computing Systems For AI Acceleration At The Extreme Edge Of The IoT

Embedding intelligence in extreme edge devices allows distilling raw data acquired from sensors into actionable information, directly on IoT end-nodes. This computing paradigm, in which end-nodes no longer depend entirely on the Cloud, offers undeniable benefits, driving a large research area (TinyML) to deploy leading Machine Learning (ML) algorithms on micro-controller class of devices. To fit the limited memory storage capability of these tiny platforms, full-precision Deep Neural Networks (DNNs) are compressed by representing their data down to byte and sub-byte formats, in the integer domain. However, the current generation of micro-controller systems can barely cope with the computing requirements of QNNs. This thesis tackles the challenge from many perspectives, presenting solutions both at software and hardware levels, exploiting parallelism, heterogeneity and software programmability to guarantee high flexibility and high energy-performance proportionality. The first contribution, PULP-NN, is an optimized software computing library for QNN inference on parallel ultra-low-power (PULP) clusters of RISC-V processors, showing one order of magnitude improvements in performance and energy efficiency, compared to current State-of-the-Art (SoA) STM32 micro-controller systems (MCUs) based on ARM Cortex-M cores. The second contribution is XpulpNN, a set of RISC-V domain specific instruction set architecture (ISA) extensions to deal with sub-byte integer arithmetic computation. The solution, including the ISA extensions and the micro-architecture to support them, achieves energy efficiency comparable with dedicated DNN accelerators and surpasses the efficiency of SoA ARM Cortex-M based MCUs, such as the low-end STM32M4 and the high-end STM32H7 devices, by up to three orders of magnitude. To overcome the Von Neumann bottleneck while guaranteeing the highest flexibility, the final contribution integrates an Analog In-Memory Computing accelerator into the PULP cluster, creating a fully programmable heterogeneous fabric that demonstrates end-to-end inference capabilities of SoA MobileNetV2 models, showing two orders of magnitude performance improvements over current SoA analog/digital solutions.

Big Data approaches as a support for precision livestock farming techniques

With the advent of new technologies it is increasingly easier to find data of different nature from even more accurate sensors that measure the most disparate physical quantities and with different methodologies. The collection of data thus becomes progressively important and takes the form of archiving, cataloging and online and offline consultation of information. Over time, the amount of data collected can become so relevant that it contains information that cannot be easily explored manually or with basic statistical techniques. The use of Big Data therefore becomes the object of more advanced investigation techniques, such as Machine Learning and Deep Learning. In this work some applications in the world of precision zootechnics and heat stress accused by dairy cows are described. Experimental Italian and German stables were involved for the training and testing of the Random Forest algorithm, obtaining a prediction of milk production depending on the microclimatic conditions of the previous days with satisfactory accuracy. Furthermore, in order to identify an objective method for identifying production drops, compared to the Wood model, typically used as an analytical model of the lactation curve, a Robust Statistics technique was used. Its application on some sample lactations and the results obtained allow us to be confident about the use of this method in the future.

Anatomical biomarkers of body composition in oncological population: application and development of artificial intelligences

In the Era of precision medicine and big medical data sharing, it is necessary to solve the work-flow of digital radiological big data in a productive and effective way. In particular, nowadays, it is possible to extract information “hidden” in digital images, in order to create diagnostic algorithms helping clinicians to set up more personalized therapies, which are in particular targets of modern oncological medicine. Digital images generated by the patient have a “texture” structure that is not visible but encrypted; it is “hidden” because it cannot be recognized by sight alone. Thanks to artificial intelligence, pre- and post-processing software and generation of mathematical calculation algorithms, we could perform a classification based on non-visible data contained in radiological images. Being able to calculate the volume of tissue body composition could lead to creating clasterized classes of patients inserted in standard morphological reference tables, based on human anatomy distinguished by gender and age, and maybe in future also by race. Furthermore, the branch of “morpho-radiology" is a useful modality to solve problems regarding personalized therapies, which is particularly needed in the oncological field. Actually oncological therapies are no longer based on generic drugs but on target personalized therapy. The lack of gender and age therapies table could be filled thanks to morpho-radiology data analysis application.