AAAyesha Abbas

Ayesha Abbas

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Ayesha Abbas has a total of 32 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Direct classification of retinal diseases from ghost imaging with deep learning

    by Ayesha Abbas, Jie Cao, Haifeng Yao, Qun Hao, Bushra Sana Idrees

    The detection of retinal diseases using conventional deep learning (DL) is reliant on fully recorded images, which restricts its use in low-resource environments where high-resolution cameras are not viable. Here, we present TRAC-Net (transformer-retina attention convolutional network) and show that retinal diseases may be directly classified from ghost imaging (GI) bucket data without the need for image reconstruction. We take a two-stage method. First, using typical retinal fundus images for training and testing, TRAC-Net achieves 98.3% accuracy, which is 5.1% better than ResNet-34. Second, we use a conventional GI configuration to physically obtain bucket measurements from retinal samples, and we use these 1D signals to train the same TRAC-Net architecture. The model yields outstanding classification results under both normal and compressed sensing conditions, demonstrating that the GI bucket signal retains diagnostically significant properties. TRAC-Net allows for scalable and hardware-efficient retinal screening especially in situations where traditional imaging is impractical by doing away with image reconstruction and working directly on bucket measurements.

  • Open Access

    Identification of Prostate Cancer Using Laser-Induced Breakdown Spectroscopy Combined with Machine Learning

    by Fiza Azam, Bushra Sana Idrees, Ejaz Khan, Amna Hameed, Rabia Nawaz, Yasir Jamil, Ayesha Abbas, Shahwal Sabir, Amna Gulzar

    Prostate cancer is a primary cause of cancer-related deaths in men which requires its rapid and accurate identification to improve treatment outcomes. This research aimed to develop an innovative, accurate and non-invasive diagnostic framework for the early detection of prostate cancer by combining laser-induced breakdown spectroscopy (LIBS) with machine learning. Prostate cancer and healthy samples were examined using LIBS to analyze their spectra. Elements like calcium, nitrogen and sodium along with the CN-band showed a higher concentration in cancer samples as compared to healthy ones. Various machine learning models were trained to accurately discriminate between cancer and healthy samples. Among several machine learning models, a trilayered neural network achieved the highest training accuracy of 85.0% and the linear SVM model achieved the highest prediction accuracy of 80.0%. Compared to other traditional methods, LIBS emerged as a robust and reliable technique. This study demonstrated the strong potential of combining LIBS with machine learning for the detection of prostate cancer. The technique not only enhanced the diagnostic accuracy but also reduced the need for invasive procedures showing promise for broader medical use.

  • Real-time target recognition with a hybrid multi-attention transformer-CNN (AttnConvNeXt) for computational ghost imaging at low sampling ratio

    by Ayesha Abbas, Rehmat Iqbal, Jie Cao, Muhammad Qasim, Tan Wang

    A hybrid multi-head attention transformer-CNN (AttnConvNeXt) model for computational ghost imaging (CGI) that can recognize targets both with and without images is presented in this paper. This unified architecture runs across several resolutions (128×128, 64×64, and 32×32) and directly analyzes raw 1D bucket measurements without reconstruction, in contrast to previous GI classifiers that were restricted to either reconstructed images or pre-processed signals. AttnConvNeXt achieves robust classification under low sampling ratios (SR=0.8) where traditional approaches fail by combining multi-head attention with convolutional layers to capture both local features and global dependencies. Our model achieves 99%–100% recognition over resolutions when used for reconstructed images, providing a high-accuracy baseline. It outperforms a 12-layer CNN by 65% when processing solely on bucket signals in image-free mode achieving 84% accuracy at SR<1. Real-time viability is demonstrated by the recognition time scaling effectively with resolution from 0.017 s/image (128×128) to 0.00056 s/signal (raw measurements). By creating the first multi-resolution, dual-mode GI recognition framework, to the best of our knowledge, this study removes the need for Fourier transforms for reconstruction-based recognition and makes it possible to use it for medical diagnostics, low-light surveillance, and scattering media.

  • Investigation of polycystic ovary syndrome using machine learning-assisted laser-induced breakdown spectroscopy

    by Rabia Nawaz, Bushra Sana Idrees, Amna Hameed, Fiza Azam, Yasir Jamil, Ayesha Abbas, Shahwal Sabir, Amna Gulzar

    Polycystic ovary syndrome (PCOS) is a significant hormonal disease that affects females. PCOS leads to complications such as infertility and metabolic issues. Investigation of PCOS is essential to reduce complications and death rate. This research facilitates the rapid detection of PCOS for effective treatment and prevention. Laser-induced breakdown spectroscopy (LIBS) was employed as a less invasive and sample-preparation-free technique for the investigation of PCOS. The elemental composition of thebsample was analyzed by atomic emission spectra. In our research, the elements such as sodium (Na), nitrogen (N), calcium (Ca), and CN- band were identified in PCOS. Machine learning models, including decision tree, neural networks, support vector machine, random forest, naive Bayes, and logistic regression classifiers, were applied to LIBS spectral data for the classification and discrimination between healthy and PCOS samples. Different models have varying training accuracy ranged from 52.9% to 82.1%, and prediction accuracy ranged from 55.0% to 93.3%. The narrow neural network attained the highest prediction accuracy of 93.3%, and the bilayer neural network attained 82.1% of training accuracy. This preliminary research of LIBS assisted with machine learning provides an alternative method to existing analytical techniques for the investigation of PCOS.


  • Efficient target recognition in ghost imaging via optimized 1D convolutional networks at a low sampling ratio

    by Ayesha Abbas, Jie Cao, Adeel Rehman, Bushra Sana Idrees, Shahwal Sabir

    The efficient use of a convolutional neural network (CNN) for object recognition in both conventional imaging and computational ghost imaging (CGI) modalities is demonstrated in this work. The model, which is trained on the Fashion-MNIST dataset, exhibits 92.91% accuracy in image-based classification; class uniqueness has a strong correlation with the performance. More importantly, the model achieves an enhanced accuracy of 97.55% in an image-free CGI framework using only compressed bucket measurements at an 80% sampling ratio. This demonstrates how it can reliably extract features directly from highly subsampled, one-dimensional data, allowing for effective, low-bandwidth optical sensing. We also investigate how resilient the architecture is to noise, and we find that a noise-augmented training approach can significantly reduce performance deterioration in high-noise situations. Thus, the suggested CNN architecture bridges the gap between conventional image processing and applied computational optics by offering a flexible and useful solution for quick, noise-resilient object recognition.

  • Cross-attention transformer enables image free target recognition in ghost imaging at ultra-low sampling rates

    by Ayesha Abbas, Jie Cao, Rehmat Iqbal, Qun Hao, Jianhua Liu, Shaotong Liu

    We propose a cross-attention mechanism combined with ghost imaging (GI) for target recognition directly from bucket measurements, eliminating the need for object imaging at extremely low sampling ratios (SR). The sparsity of measurements at ultra-low SR poses a challenge for the traditional recognition methods. Inspired by the cross-attention mechanism’s ability to highlight relevant features within data, our approach focuses on the most relevant input features within bucket measurements while disregarding noisy and irrelevant information. Our cross-attention DL model trained on GI measurements at 0.2 and 0.1 SRs achieves high recognition accuracy of 99% and 96% respectively. The cross-attention model remains robust against data augmentation and capable of classifying multi-class targets without any need for imaging, and hence eliminates the reconstruction time at ultra-low SRs maintained at 0.1 and 0.2. Experimental and simulated results validate the efficacy of our model, demonstrating the importance of image-free target recognition in autonomous vehicles, remote sensing, medical diagnostics, defense, and industrial inspection, where efficient recognition with sparse data is critical.


  • Fourier domain ghost imaging with adaptive enhancement for scattering media

    by Rehmat Iqbal, Cao Jie, Ayesha Abbas, Kai Lin, Qun Hao

    Remote-sensing applications within atmospheric scattering environments pose substantial challenges for traditional optical imaging systems owing to photon scattering, signal attenuation, and noise-induced degradation. This paper presents a ghost imaging technique operating in the Fourier domain, integrated with adaptive multistage post-processing enhancement algorithms to improve image reconstruction fidelity in the presence of scattering media. The experimental framework utilizes a digital micromirror device (DMD) to generate structured Fourier basis patterns in the reference arm, while the signal arm traverses controllable scattering media simulated using a precision-positioned ground glass (GG) diffuser. Image reconstruction is achieved through frequency-domain correlation processing employing a three-step phase-shifting algorithm, while reconstructed image degradation is mitigated through a multi-stage enhancement pipeline incorporating adaptive denoising, deblurring, and contrast optimization algorithms. Experimental validation demonstrates that the proposed methodology achieves recognizable image reconstruction using 25% of the Fourier coefficient set under high-scattering conditions, with enhancement algorithms providing up to 30% improvement in peak signal-to-noise ratio (PSNR) relative to unprocessed reconstructions. This investigation reveals fundamental limitations in dense scattering regimes and provides insights into the computational trade-offs inherent in ghost imaging systems under challenging environmental conditions.


  • Ablative laser propulsion, the propellants and measuring parameters: A brief review

    by Ayesha Abbas, Syeda Tehreem Iqbal, Yasir Jamil

    Keeping in view the weightage of electric propulsion over chemical propulsion, materials from metals to polymers and liquid (water) have been tested as propellants in ablative laser propulsion. This emerging propulsion technique can be widely used for aerospace applications like debris removal in the range of cm, pointing micro and nano satellites, laser micro thrusters for spacecraft attitude, and orbit control. Laser propulsion can become a less expensive alternative to chemical propulsion. In this review, we compile the work done in ablative laser propulsion and different modes of propulsion along with the efficiency of different propellants. We summarize the optimized propulsive parameters with solid propellants and liquid propellants along with the efficiencies and theories of laser thrusters with optimized specific impulses. The article provides precise developments done in the field of ablative laser propulsion and deep insights into the analysis done between the different propellants used recently in ablative laser propulsion.


  • Open Access

    Experimenting on semiconductors using ablative laser propulsion to investigate propulsion parameters

    by Ayesha Abbas, Muhammad Qaiser Zakaria, Syeda Tehreem Iqbal, Yasir Jamil

    A laser can be used to propel distant objects. The ability of silicon to produce thrust and its propulsive parameters are needed to be studied in ablative laser propulsion (ALP). In this work, pure silicon and silicon doped with indium were subjected to ALP to achieve the momentum coupling coefficient and specific impulse to investigate its worth as a propellant. The experiment was conducted using the Nd:YAG laser (Quantel Brilliant) operating at fundamental harmonic (λ = 1064 nm and 5 ns pulse duration). In the given range of fluence, 1 × 105–5 × 105 J/m2, no significant difference among both samples for values of momentum coupling coefficient (Cm) and specific impulse (Isp) is observed; for instance, both propellants follow a decreasing trend for Cm. However, maximum enhancement for Cm and Isp is observed with a cavity aspect ratio of one. Cm and Isp are enhanced about 1.75 and 4.5 times, respectively, for pure silicon. The external cavity does not have any impact on Cm values for indium-doped silicon while values of Isp showed considerable enhancement.