AI Applications in Radiology
Expert-defined terms from the Professional Certificate in AI in Medical Imaging course at London School of International Business. Free to read, free to share, paired with a globally recognised certification pathway.
**Artificial Intelligence (AI)** #
**Artificial Intelligence (AI)**
AI refers to the simulation of human intelligence in machines that are programme… #
The subfields of AI include machine learning, deep learning, and natural language processing.
**Machine Learning (ML)** #
**Machine Learning (ML)**
ML is a subset of AI that enables machines to learn and improve from experience… #
It uses algorithms to analyze data, identify patterns, and make decisions with minimal human intervention.
**Deep Learning (DL)** #
**Deep Learning (DL)**
DL is a subset of ML that uses artificial neural networks with many layers (henc… #
It can process large volumes of unstructured data, such as images and sound, and is commonly used in applications such as image recognition and natural language processing.
**Natural Language Processing (NLP)** #
**Natural Language Processing (NLP)**
NLP is a subset of AI that deals with the interaction between computers and huma… #
It enables machines to understand, interpret, and generate human language in a valuable way.
**Convolutional Neural Networks (CNNs)** #
**Convolutional Neural Networks (CNNs)**
CNNs are a class of deep learning models that are designed to process data with… #
They are composed of multiple layers, including convolutional layers, pooling layers, and fully connected layers.
**Recurrent Neural Networks (RNNs)** #
**Recurrent Neural Networks (RNNs)**
RNNs are a class of deep learning models that are designed to process sequential… #
They have a feedback loop that allows information from previous time steps to influence the current step.
**Generative Adversarial Networks (GANs)** #
**Generative Adversarial Networks (GANs)**
GANs are a class of deep learning models that consist of two components #
a generator and a discriminator. The generator creates new data instances, while the discriminator evaluates the authenticity of the generated data.
**Fully Convolutional Networks (FCNs)** #
**Fully Convolutional Networks (FCNs)**
FCNs are a class of deep learning models that are composed entirely of convoluti… #
They are commonly used for image segmentation tasks.
**Transfer Learning** #
**Transfer Learning**
Transfer learning is a technique in ML where a pre #
trained model is used as a starting point for a new task. It enables models to leverage the knowledge and features learned from a related task, reducing the amount of data and computation required.
**Data Augmentation** #
**Data Augmentation**
Data augmentation is a technique in ML where artificial data is generated by app… #
It enables models to learn from a more diverse set of data, improving their generalization ability.
**Active Learning** #
**Active Learning**
Active learning is a technique in ML where the model selects the most informativ… #
It enables models to learn more efficiently and effectively, reducing the amount of labeling required.
**Fully Supervised Learning** #
**Fully Supervised Learning**
Fully supervised learning is a type of ML where the model is trained on labeled… #
It is commonly used for tasks such as image classification and object detection.
**Semi #
Supervised Learning**
Semi #
supervised learning is a type of ML where the model is trained on a combination of labeled and unlabeled data. It is commonly used for tasks where labeled data is scarce or expensive to obtain.
**Unsupervised Learning** #
**Unsupervised Learning**
Unsupervised learning is a type of ML where the model is trained on unlabeled da… #
It is commonly used for tasks such as clustering and dimensionality reduction.
**Reinforcement Learning** #
**Reinforcement Learning**
Reinforcement learning is a type of ML where the model learns by interacting wit… #
It is commonly used for tasks such as game playing and robotics.
**Precision** #
**Precision**
Precision is a measure of the correctness of a model's predictions #
It is defined as the number of true positives divided by the sum of true positives and false positives.
**Recall** #
**Recall**
Recall is a measure of the completeness of a model's predictions #
It is defined as the number of true positives divided by the sum of true positives and false negatives.
**F1 Score** #
**F1 Score**
The F1 score is a measure of a model's accuracy that balances precision and reca… #
It is defined as the harmonic mean of precision and recall.
**Confusion Matrix** #
**Confusion Matrix**
A confusion matrix is a table that summarizes the performance of a model's predi… #
It shows the number of true positives, true negatives, false positives, and false negatives.
**Interpretability** #
**Interpretability**
Interpretability is the ability of a model to explain its decisions in a way tha… #
It is important in radiology applications to ensure that the model's decisions can be trusted and validated.
**Explainability** #
**Explainability**
Explainability is the ability of a model to provide insights into how it arrived… #
It is important in radiology applications to ensure that the model's decisions can be understood and validated.
**Feature Engineering** #
**Feature Engineering**
Feature engineering is the process of selecting and transforming raw data into f… #
It is an important step in ML to ensure that the model has access to relevant and informative data.
**Overfitting** #
**Overfitting**
Overfitting is a common problem in ML where a model learns the training data too… #
It can be addressed by using regularization techniques, such as dropout and weight decay.
**Underfitting** #
**Underfitting**
Underfitting is a common problem in ML where a model fails to learn the underlyi… #
It can be addressed by using more complex models, adding more features, or collecting more data.
**Cross #
Validation**
Cross #
validation is a technique in ML where the data is split into multiple folds, and the model is trained and evaluated on each fold. It enables the model's performance to be evaluated more accurately, reducing the risk of overfitting.
**Batch Normalization** #
**Batch Normalization**
Batch normalization is a technique in DL where the inputs to each layer are norm… #
It enables models to converge faster and achieve better accuracy.
**Activation Function** #
**Activation Function**
An activation function is a function that is applied to the output of a neural n… #
Common activation functions include the sigmoid, tanh, and ReLU functions.
**Loss Function** #
**Loss Function**
A loss function is a function that measures the difference between the model's p… #
It is used to optimize the model's parameters during training.
**Optimization Algorithm** #
**Optimization Algorithm**
An optimization algorithm is a method for updating the model's parameters to min… #
Common optimization algorithms include stochastic gradient descent, Adam, and RMSprop.
**Early Stopping** #
**Early Stopping**
Early stopping is a technique in DL where the training process is stopped when t… #
It helps to prevent overfitting and improve the model's generalization ability.
**Regularization** #
**Regularization**
Regularization is a technique in DL that is used to prevent overfitting by addin… #
Common regularization techniques include L1 and L2 regularization, dropout, and weight decay.
**Data Preprocessing** #
**Data Preprocessing**
Data preprocessing is the process of cleaning, normalizing, and transforming raw… #
It is an important step in ML to ensure that the data is of high quality and free of bias.
**Data Quality** #
**Data Quality**
Data quality is a measure of the accuracy, completeness, and consistency of the… #
It is important in radiology applications to ensure that the data is of high quality and free of errors.
**Data Bias** #
**Data Bias**
Data bias is a systematic error in the data that can lead to biased or inaccurat… #
It is important in radiology applications to ensure that the data is free of bias and representative of the population.
**Data Security** #
**Data Security**
Data security is the protection of data from unauthorized access, use, disclosur… #
It is important in radiology applications to ensure that patient data is kept confidential and secure.
**Data Privacy** #
**Data Privacy**
Data privacy is the right of individuals to control the collection, use, and dis… #
It is important in radiology applications to ensure that patient data is used only for the intended purpose and with their consent.
**Data Integrity** #
**Data Integrity**
Data integrity is the assurance that the data is accurate, complete, and consist… #
It is important in radiology applications to ensure that the data is trustworthy and reliable.
**Data Governance** #
**Data Governance**
Data governance is the overall management and control of data throughout its lif… #
It includes the development