-
Comparison of fine-tuning strategies for transfer learning in medical image classification
Authors:
Ana Davila,
Jacinto Colan,
Yasuhisa Hasegawa
Abstract:
In the context of medical imaging and machine learning, one of the most pressing challenges is the effective adaptation of pre-trained models to specialized medical contexts. Despite the availability of advanced pre-trained models, their direct application to the highly specialized and diverse field of medical imaging often falls short due to the unique characteristics of medical data. This study…
▽ More
In the context of medical imaging and machine learning, one of the most pressing challenges is the effective adaptation of pre-trained models to specialized medical contexts. Despite the availability of advanced pre-trained models, their direct application to the highly specialized and diverse field of medical imaging often falls short due to the unique characteristics of medical data. This study provides a comprehensive analysis on the performance of various fine-tuning methods applied to pre-trained models across a spectrum of medical imaging domains, including X-ray, MRI, Histology, Dermoscopy, and Endoscopic surgery. We evaluated eight fine-tuning strategies, including standard techniques such as fine-tuning all layers or fine-tuning only the classifier layers, alongside methods such as gradually unfreezing layers, regularization based fine-tuning and adaptive learning rates. We selected three well-established CNN architectures (ResNet-50, DenseNet-121, and VGG-19) to cover a range of learning and feature extraction scenarios. Although our results indicate that the efficacy of these fine-tuning methods significantly varies depending on both the architecture and the medical imaging type, strategies such as combining Linear Probing with Full Fine-tuning resulted in notable improvements in over 50% of the evaluated cases, demonstrating general effectiveness across medical domains. Moreover, Auto-RGN, which dynamically adjusts learning rates, led to performance enhancements of up to 11% for specific modalities. Additionally, the DenseNet architecture showed more pronounced benefits from alternative fine-tuning approaches compared to traditional full fine-tuning. This work not only provides valuable insights for optimizing pre-trained models in medical image analysis but also suggests the potential for future research into more advanced architectures and fine-tuning methods.
△ Less
Submitted 14 June, 2024;
originally announced June 2024.
-
Manipulability maximization in constrained inverse kinematics of surgical robots
Authors:
Jacinto Colan,
Ana Davila,
Yasuhisa Hasegawa
Abstract:
In robot-assisted minimally invasive surgery (RMIS), inverse kinematics (IK) must satisfy a remote center of motion (RCM) constraint to prevent tissue damage at the incision point. However, most of existing IK methods do not account for the trade-offs between the RCM constraint and other objectives such as joint limits, task performance and manipulability optimization. This paper presents a novel…
▽ More
In robot-assisted minimally invasive surgery (RMIS), inverse kinematics (IK) must satisfy a remote center of motion (RCM) constraint to prevent tissue damage at the incision point. However, most of existing IK methods do not account for the trade-offs between the RCM constraint and other objectives such as joint limits, task performance and manipulability optimization. This paper presents a novel method for manipulability maximization in constrained IK of surgical robots, which optimizes the robot's dexterity while respecting the RCM constraint and joint limits. Our method uses a hierarchical quadratic programming (HQP) framework that solves a series of quadratic programs with different priority levels. We evaluate our method in simulation on a 6D path tracking task for constrained and unconstrained IK scenarios for redundant kinematic chains. Our results show that our method enhances the manipulability index for all cases, with an important increase of more than 100% when a large number of degrees of freedom are available. The average computation time for solving the IK problems was under 1ms, making it suitable for real-time robot control. Our method offers a novel and effective solution to the constrained IK problem in RMIS applications.
△ Less
Submitted 14 June, 2024;
originally announced June 2024.
-
Task segmentation based on transition state clustering for surgical robot assistance
Authors:
Yutaro Yamada,
Jacinto Colan,
Ana Davila,
Yasuhisa Hasegawa
Abstract:
Understanding surgical tasks represents an important challenge for autonomy in surgical robotic systems. To achieve this, we propose an online task segmentation framework that uses hierarchical transition state clustering to activate predefined robot assistance. Our approach involves performing a first clustering on visual features and a subsequent clustering on robot kinematic features for each v…
▽ More
Understanding surgical tasks represents an important challenge for autonomy in surgical robotic systems. To achieve this, we propose an online task segmentation framework that uses hierarchical transition state clustering to activate predefined robot assistance. Our approach involves performing a first clustering on visual features and a subsequent clustering on robot kinematic features for each visual cluster. This enables to capture relevant task transition information on each modality independently. The approach is implemented for a pick-and-place task commonly found in surgical training. The validation of the transition segmentation showed high accuracy and fast computation time. We have integrated the transition recognition module with predefined robot-assisted tool positioning. The complete framework has shown benefits in reducing task completion time and cognitive workload.
△ Less
Submitted 14 June, 2024;
originally announced June 2024.
-
Constrained Motion Planning for a Robotic Endoscope Holder based on Hierarchical Quadratic Programming
Authors:
Jacinto Colan,
Ana Davila,
Yasuhisa Hasegawa
Abstract:
Minimally Invasive Surgeries (MIS) are challenging for surgeons due to the limited field of view and constrained range of motion imposed by narrow access ports. These challenges can be addressed by robot-assisted endoscope systems which provide precise and stabilized positioning, as well as constrained and smooth motion control of the endoscope. In this work, we propose an online hierarchical opti…
▽ More
Minimally Invasive Surgeries (MIS) are challenging for surgeons due to the limited field of view and constrained range of motion imposed by narrow access ports. These challenges can be addressed by robot-assisted endoscope systems which provide precise and stabilized positioning, as well as constrained and smooth motion control of the endoscope. In this work, we propose an online hierarchical optimization framework for visual servoing control of the endoscope in MIS. The framework prioritizes maintaining a remote-center-of-motion (RCM) constraint to prevent tissue damage, while a visual tracking task is defined as a secondary task to enable autonomous tracking of visual features of interest. We validated our approach using a 6-DOF Denso VS050 manipulator and achieved optimization solving times under 0.4 ms and maximum RCM deviation of approximately 0.4 mm. Our results demonstrate the effectiveness of the proposed approach in addressing the constrained motion planning challenges of MIS, enabling precise and autonomous endoscope positioning and visual tracking.
△ Less
Submitted 14 June, 2024;
originally announced June 2024.
-
mldr.resampling: Efficient Reference Implementations of Multilabel Resampling Algorithms
Authors:
Antonio J. Rivera,
Miguel A. Dávila,
David Elizondo,
María J. del Jesus,
Francisco Charte
Abstract:
Resampling algorithms are a useful approach to deal with imbalanced learning in multilabel scenarios. These methods have to deal with singularities in the multilabel data, such as the occurrence of frequent and infrequent labels in the same instance. Implementations of these methods are sometimes limited to the pseudocode provided by their authors in a paper. This Original Software Publication pre…
▽ More
Resampling algorithms are a useful approach to deal with imbalanced learning in multilabel scenarios. These methods have to deal with singularities in the multilabel data, such as the occurrence of frequent and infrequent labels in the same instance. Implementations of these methods are sometimes limited to the pseudocode provided by their authors in a paper. This Original Software Publication presents mldr.resampling, a software package that provides reference implementations for eleven multilabel resampling methods, with an emphasis on efficiency since these algorithms are usually time-consuming.
△ Less
Submitted 30 May, 2023; v1 submitted 26 May, 2023;
originally announced May 2023.
-
"We need to do more ... I need to do more": Augmenting Digital Media Consumption via Critical Reflection to Increase Compassion and Promote Prosocial Attitudes and Behaviors
Authors:
Ken Jen Lee,
Adrian Davila,
Hanlin Cheng,
Joslin Goh,
Elizabeth Nilsen,
Edith Law
Abstract:
Much HCI research on prompting prosocial behaviors focuses on methods for increasing empathy. However, increased empathy may have unintended negative consequences. Our work offers an alternative solution that encourages critical reflection for nurturing compassion, which involves motivation and action to help others. In a between-subject experiment, participants (N=60) viewed a climate change docu…
▽ More
Much HCI research on prompting prosocial behaviors focuses on methods for increasing empathy. However, increased empathy may have unintended negative consequences. Our work offers an alternative solution that encourages critical reflection for nurturing compassion, which involves motivation and action to help others. In a between-subject experiment, participants (N=60) viewed a climate change documentary while receiving no prompts (CON), reflective prompts to focus on their emotions (RE) or surprises (RS). State compassion, critical reflection, and motivation to act or learn were measured at the end of the session (post-video) and two weeks later (follow-up). Despite participants' condition not affecting compassion, critical reflection was positively correlated with post-video state compassion. RE and RS participants demonstrated deeper reflection and reported higher motivation to learn post-video, and more prosocial behavioral changes during follow-up. RS participants reported better follow-up recall than RE participants. We conclude by discussing implications on designing technology to support compassion and longer-term critical reflection.
△ Less
Submitted 7 May, 2023;
originally announced May 2023.
-
Habitability Models for Astrobiology
Authors:
Abel Méndez,
Edgard E. Rivera-Valentín,
Dirk Schulze-Makuch,
Justin Filiberto,
Ramses M. Ramírez,
Tana Wood,
Alfonso Dávila,
Chris McKay,
Kevin N. Ortiz Ceballos,
Marcos Jusino-Maldonado,
Nicole J. Torres-Santiago,
Guillermo Nery,
René Heller,
Paul K. Byrne,
Michael J. Malaska,
Erica Nathan,
Marta F. Simões,
André Antunes,
Jesús Martínez-Frías,
Ludmila Carone,
Noam R. Izenberg,
Dimitra Atri,
Humberto I. Carvajal Chitty,
Priscilla Nowajewski-Barra,
Frances Rivera-Hernández
, et al. (9 additional authors not shown)
Abstract:
Habitability has been generally defined as the capability of an environment to support life. Ecologists have been using Habitat Suitability Models (HSMs) for more than four decades to study the habitability of Earth from local to global scales. Astrobiologists have been proposing different habitability models for some time, with little integration and consistency among them, being different in fun…
▽ More
Habitability has been generally defined as the capability of an environment to support life. Ecologists have been using Habitat Suitability Models (HSMs) for more than four decades to study the habitability of Earth from local to global scales. Astrobiologists have been proposing different habitability models for some time, with little integration and consistency among them, being different in function to those used by ecologists. Habitability models are not only used to determine if environments are habitable or not, but they also are used to characterize what key factors are responsible for the gradual transition from low to high habitability states. Here we review and compare some of the different models used by ecologists and astrobiologists and suggest how they could be integrated into new habitability standards. Such standards will help to improve the comparison and characterization of potentially habitable environments, prioritize target selections, and study correlations between habitability and biosignatures. Habitability models are the foundation of planetary habitability science and the synergy between ecologists and astrobiologists is necessary to expand our understanding of the habitability of Earth, the Solar System, and extrasolar planets.
△ Less
Submitted 11 August, 2021;
originally announced August 2021.
-
Habitability Models for Planetary Sciences
Authors:
Abel Méndez,
Edgard G. Rivera-Valentín,
Dirk Schulze-Makuch,
Justin Filiberto,
Ramses Ramírez,
Tana E. Wood,
Alfonso Dávila,
Chris McKay,
Kevin Ortiz Ceballos,
Marcos Jusino-Maldonado,
Guillermo Nery,
René Heller,
Paul Byrne,
Michael J. Malaska,
Erica Nathan,
Marta Filipa Simões,
André Antunes,
Jesús Martínez-Frías,
Ludmila Carone,
Noam R. Izenberg,
Dimitra Atri,
Humberto Itic Carvajal Chitty,
Priscilla Nowajewski-Barra,
Frances Rivera-Hernández,
Corine Brown
, et al. (10 additional authors not shown)
Abstract:
Habitability has been generally defined as the capability of an environment to support life. Ecologists have been using Habitat Suitability Models (HSMs) for more than four decades to study the habitability of Earth from local to global scales. Astrobiologists have been proposing different habitability models for some time, with little integration and consistency between them and different in func…
▽ More
Habitability has been generally defined as the capability of an environment to support life. Ecologists have been using Habitat Suitability Models (HSMs) for more than four decades to study the habitability of Earth from local to global scales. Astrobiologists have been proposing different habitability models for some time, with little integration and consistency between them and different in function to those used by ecologists. In this white paper, we suggest a mass-energy habitability model as an example of how to adapt and expand the models used by ecologists to the astrobiology field. We propose to implement these models into a NASA Habitability Standard (NHS) to standardize the habitability objectives of planetary missions. These standards will help to compare and characterize potentially habitable environments, prioritize target selections, and study correlations between habitability and biosignatures. Habitability models are the foundation of planetary habitability science. The synergy between the methods used by ecologists and astrobiologists will help to integrate and expand our understanding of the habitability of Earth, the Solar System, and exoplanets.
△ Less
Submitted 14 July, 2020; v1 submitted 10 July, 2020;
originally announced July 2020.
-
Texture Interpolation for Probing Visual Perception
Authors:
Jonathan Vacher,
Aida Davila,
Adam Kohn,
Ruben Coen-Cagli
Abstract:
Texture synthesis models are important tools for understanding visual processing. In particular, statistical approaches based on neurally relevant features have been instrumental in understanding aspects of visual perception and of neural coding. New deep learning-based approaches further improve the quality of synthetic textures. Yet, it is still unclear why deep texture synthesis performs so wel…
▽ More
Texture synthesis models are important tools for understanding visual processing. In particular, statistical approaches based on neurally relevant features have been instrumental in understanding aspects of visual perception and of neural coding. New deep learning-based approaches further improve the quality of synthetic textures. Yet, it is still unclear why deep texture synthesis performs so well, and applications of this new framework to probe visual perception are scarce. Here, we show that distributions of deep convolutional neural network (CNN) activations of a texture are well described by elliptical distributions and therefore, following optimal transport theory, constraining their mean and covariance is sufficient to generate new texture samples. Then, we propose the natural geodesics (ie the shortest path between two points) arising with the optimal transport metric to interpolate between arbitrary textures. Compared to other CNN-based approaches, our interpolation method appears to match more closely the geometry of texture perception, and our mathematical framework is better suited to study its statistical nature. We apply our method by measuring the perceptual scale associated to the interpolation parameter in human observers, and the neural sensitivity of different areas of visual cortex in macaque monkeys.
△ Less
Submitted 22 October, 2020; v1 submitted 5 June, 2020;
originally announced June 2020.
-
Limits on Observation in Quantum Gravity and Black Holes
Authors:
George Alexander Davila
Abstract:
We discuss how the bounds on observation associated with the Planck units would affect an observers perception of a black hole. By simply imposing Planck scale quantities as the lower bounds for length, time, and mass of black hole formation, interesting insights into the nature of black holes can be gained. We also see the emergence of a Planck-scale mass that plays an important role in the obser…
▽ More
We discuss how the bounds on observation associated with the Planck units would affect an observers perception of a black hole. By simply imposing Planck scale quantities as the lower bounds for length, time, and mass of black hole formation, interesting insights into the nature of black holes can be gained. We also see the emergence of a Planck-scale mass that plays an important role in the observation of black holes and the emergence of a new mechanism for virtual black hole formation.
△ Less
Submitted 7 October, 2014; v1 submitted 15 September, 2014;
originally announced September 2014.
-
Proton/Pion ratios in Δφwith respect to a jet in sqrt{s_{NN}} =200 GeV Au+Au collisions at STAR
Authors:
Alán Dávila
Abstract:
Inclusive Proton/pion ratios show an enhancement at intermediate transverse momentum (P_{T} ~ 1.5 - 4.0 GeV/c) in central sqrt{s_{NN}} = 200 GeV Au+Au collisions compared to peripheral Au+Au and p+p collisions. This effect suggests a production mechanism, different from fragmentation, which is consistent with coalescence and recombination models. A high E_{T} trigger particle selects a surface-bia…
▽ More
Inclusive Proton/pion ratios show an enhancement at intermediate transverse momentum (P_{T} ~ 1.5 - 4.0 GeV/c) in central sqrt{s_{NN}} = 200 GeV Au+Au collisions compared to peripheral Au+Au and p+p collisions. This effect suggests a production mechanism, different from fragmentation, which is consistent with coalescence and recombination models. A high E_{T} trigger particle selects a surface-biased jet, which is measured to have a similar p_{T} distribution as a p+p jet. This jet is used to enhance the quenching effects of the recoiling, medium traversing one. We reconstruct the trigger jet using the FASTJET algorithm, with a E_{T} (p_{T}) cut of 3.0 GeV(/c) on the towers(tracks) in order to reduce the heavy-ion collision background. The particle identification of tracks with p_{T} up to ~ 2.8 GeV/c is obtained by taking advantage of STAR TOF and TPC detectors with full azimuth coverage. Correlations in Δφbetween jets and identified hadrons are presented, and the particle ratios in different regions of azimuth are measured. Particle ratios associated with the trigger jet vs. the recoil jet, and comparisons to inclusive particle ratios can help to distinguish between jet-related (vacuum and medium-modified) and bulk-related contributions to the ratios enhancement.
△ Less
Submitted 21 September, 2012; v1 submitted 30 July, 2012;
originally announced July 2012.
-
A study of structural properties on profiles HMMs
Authors:
Juliana S Bernardes,
Alberto Davila,
Vitor Santos Costa,
Gerson Zaverucha
Abstract:
Motivation: Profile hidden Markov Models (pHMMs) are a popular and very useful tool in the detection of the remote homologue protein families. Unfortunately, their performance is not always satisfactory when proteins are in the 'twilight zone'. We present HMMER-STRUCT, a model construction algorithm and tool that tries to improve pHMM performance by using structural information while training pH…
▽ More
Motivation: Profile hidden Markov Models (pHMMs) are a popular and very useful tool in the detection of the remote homologue protein families. Unfortunately, their performance is not always satisfactory when proteins are in the 'twilight zone'. We present HMMER-STRUCT, a model construction algorithm and tool that tries to improve pHMM performance by using structural information while training pHMMs. As a first step, HMMER-STRUCT constructs a set of pHMMs. Each pHMM is constructed by weighting each residue in an aligned protein according to a specific structural property of the residue. Properties used were primary, secondary and tertiary structures, accessibility and packing. HMMER-STRUCT then prioritizes the results by voting. Results: We used the SCOP database to perform our experiments. Throughout, we apply leave-one-family-out cross-validation over protein superfamilies. First, we used the MAMMOTH-mult structural aligner to align the training set proteins. Then, we performed two sets of experiments. In a first experiment, we compared structure weighted models against standard pHMMs and against each other. In a second experiment, we compared the voting model against individual pHMMs. We compare method performance through ROC curves and through Precision/Recall curves, and assess significance through the paired two tailed t-test. Our results show significant performance improvements of all structurally weighted models over default HMMER, and a significant improvement in sensitivity of the combined models over both the original model and the structurally weighted models.
△ Less
Submitted 11 December, 2008; v1 submitted 16 April, 2007;
originally announced April 2007.
-
Geometrical aspects of isoscaling
Authors:
A. Dávila,
C. Escudero,
J. A. López,
C. O. Dorso
Abstract:
The property of isoscaling in nuclear fragmentation is studied using a simple bond percolation model with ``isospin'' added as an extra degree of freedom. It is shown analytically, first, that isoscaling is expected to exist in such a simple model with the only assumption of fair sampling with homogeneous probabilities. Second, numerical percolations of hundreds of thousands of grids of differen…
▽ More
The property of isoscaling in nuclear fragmentation is studied using a simple bond percolation model with ``isospin'' added as an extra degree of freedom. It is shown analytically, first, that isoscaling is expected to exist in such a simple model with the only assumption of fair sampling with homogeneous probabilities. Second, numerical percolations of hundreds of thousands of grids of different sizes and with different $N$ to $Z$ ratios confirm this prediction with remarkable agreement. It is thus concluded that isoscaling emerges from the simple assumption of fair sampling with homogeneous probabilities, a requirement which, if put in the nomenclature of the minimum information theory, translates simply into the existence of equiprobable configurations in maximum entropy states.
△ Less
Submitted 12 April, 2005;
originally announced April 2005.
