roboticsfrontier مؤسسه پژوهش‌های مرزی رباتیک
Menu

Research / Sheet T-02 — Program 02

Perception for manipulation

Recognising states, events and structures in physiological signals, speech and medical images.

T2 / 01

Problem statement

Physiological recordings, speech and medical images contain patterns that matter: a sleep arousal, an approaching seizure, an emotional response, the course of a coronary artery. Some are brief or rare, and marking them by hand, as in sleep scoring, is slow work that needs specialist knowledge. The program studies how a machine can perceive them reliably: which representations of EEG, ECG-derived interval series, speech, X-ray angiograms and paired MRI and PET scans expose the relevant structure, and which learning methods then classify or segment it accurately, at modest computational cost and with few false alarms.

T2 / 02

Scientific challenge

  • Sleep arousals are brief, rare events in long EEG recordings, so detectors must cope with class imbalance and keep false positives low.

  • ECG interval series give only indirect evidence of an approaching seizure or an emotional response, reflected through the autonomic nervous system.

  • Medical images raise other problems: separating coronary vessels from background in X-ray angiograms, and registering MRI with PET before fusing their features.

T2 / 03

Approach

  1. Recast signals as geometric or time-frequency representations: Poincaré plots of ECG intervals, spectrogram images of EEG windows, phase-space tensors of speech.

  2. Classify with deep convolutional networks, some with deformable multiscale attention, or with feature-based classifiers such as SVMs; segment vessels by density-based clustering.

  3. Fuse complementary sources, EEG with ECG and MRI with PET, and keep models economical through single-lead inputs, transfer learning and autoencoders.

T2 / 04

Related publications

  • 2026

    A Novel Non-Real-Time Algorithm for Epileptic Seizure Prediction Using Features Extracted from Multiple Time Series Derived from the Two-Dimensional Poincaré Plot of RR Intervals

    Pardis Goodarzi, Keivan Maghooli, Nader Jafarnia Dabanloo, Fardad Farokhi · Iranian Journal of Science and Technology, Transactions of Electrical Engineering · Department of Biomedical Engineering, CT.C., Islamic Azad University, Tehran, Iran · DOI:10.1007/s40998-026-01023-4

    DOI ↗ BibTeX

    journal Prior affiliation
  • 2025

    A Fuzzy Cognitive Map-based Framework for Alzheimer's Disease Diagnosis Using Multimodal Magnetic Resonance Imaging-Positron Emission Tomography Registration

    Seyed Assef Mahdavi, Keivan Maghooli, Fardad Farokhi · Journal of Medical Signals & Sensors, vol. 15, art. 31 · Department of Biomedical Engineering, Central Tehran Branch, Islamic Azad University, Tehran, Iran · DOI:10.4103/jmss.jmss_3_25

    DOI ↗ BibTeX

    journal Prior affiliation
  • 2025

    An approach to arousal disorder classification using deformable convolution and adaptive multiscale features in EEG signals

    Andia Foroughi, Fardad Farokhi, Fereidoun Nowshiravan Rahatabad, Alireza Kashaninia · Brain Research Bulletin, vol. 230, art. 111468 · Department of Biomedical Engineering, Central Tehran Branch, Islamic Azad University, Tehran, Iran · DOI:10.1016/j.brainresbull.2025.111468

    DOI ↗ BibTeX

    journal Prior affiliation
  • 2025

    Segmentation of coronary arteries from X-ray angiographic images using density based spatial clustering of applications with noise (DBSCAN)

    Kamran Mardani, Keivan Maghooli, Fardad Farokhi · Biomedical Signal Processing and Control · Department of Biomedical Engineering, Central Tehran Branch, Islamic Azad University, Tehran, Iran · DOI:10.1016/j.bspc.2024.107175

    DOI ↗ BibTeX

    journal Prior affiliation
  • 2023

    A 3D Tensor Representation of Speech and 3D Convolutional Neural Network for Emotion Recognition

    Mohammad Reza Falahzadeh, Fardad Farokhi, Ali Harimi, Reza Sabbaghi-Nadooshan · Circuits, Systems, and Signal Processing · Department of Biomedical Engineering, Central Tehran Branch, Islamic Azad University, Tehran, Iran · DOI:10.1007/s00034-023-02315-4

    DOI ↗ BibTeX

    journal Prior affiliation
  • 2023

    Deep convolutional architecture-based hybrid learning for sleep arousal events detection through single-lead EEG signals

    Andia Foroughi, Fardad Farokhi, Fereidoun Nowshiravan Rahatabad, Alireza Kashaninia · Brain and Behavior, vol. 13, no. 6, art. e3028 · Department of Electrical Engineering, Central Tehran Branch Islamic Azad University, Tehran, Iran · DOI:10.1002/brb3.3028

    DOI ↗ BibTeX

    journal Prior affiliation
  • 2022

    Deep Convolutional Neural Network and Gray Wolf Optimization Algorithm for Speech Emotion Recognition

    Mohammad Reza Falahzadeh, Fardad Farokhi, Ali Harimi, Reza Sabbaghi-Nadooshan · Circuits, Systems, and Signal Processing · Department of Biomedical Engineering, Central Tehran Branch, Islamic Azad University, Tehran, Iran · DOI:10.1007/s00034-022-02130-3

    DOI ↗ BibTeX

    journal Prior affiliation
  • 2020

    A new emotion detection algorithm using extracted features of the different time-series generated from ST intervals Poincaré map

    Maryam Baghizadeh, Keivan Maghooli, Fardad Farokhi, Nader Jafarnia Dabanloo · Biomedical Signal Processing and Control, art. 101902 · Department of Biomedical Engineering, Central Tehran Branch, Islamic Azad University, Tehran, Iran · DOI:10.1016/j.bspc.2020.101902

    DOI ↗ BibTeX

    journal Prior affiliation

All publications

T2 / 05

People

Faculty on this program

Students on this program

Students — none listed yet

A student is named here only once their consent is recorded.

Open positions

Open positions — none posted on this program

Positions are posted with their degree and funding terms when they open. Work with us is the route in the meantime.

Have a problem in perception for manipulation — academic or industrial? The scoping conversation costs nothing.

Discuss a collaboration