This article explores the methods and datasets used to build a benchmark for content-based image retrieval (CBIR) in medical imaging. It examines vector databases, the challenges of large-scale similarity search, and indexing techniques such as flat search, Locality Sensitive Hashing (LSH), and Hierarchical Navigable Small World (HNSW). The Facebook AI Similarity Search (FAISS) library is used to implement efficient approximate nearest neighbor (ANN) search. Using the TotalSegmentator dataset of over 1,200 CT volumes, embeddings were extracted slice-by-slice and indexed, enabling rapid, metadata-free retrieval across more than 290,000 image embeddings.This article explores the methods and datasets used to build a benchmark for content-based image retrieval (CBIR) in medical imaging. It examines vector databases, the challenges of large-scale similarity search, and indexing techniques such as flat search, Locality Sensitive Hashing (LSH), and Hierarchical Navigable Small World (HNSW). The Facebook AI Similarity Search (FAISS) library is used to implement efficient approximate nearest neighbor (ANN) search. Using the TotalSegmentator dataset of over 1,200 CT volumes, embeddings were extracted slice-by-slice and indexed, enabling rapid, metadata-free retrieval across more than 290,000 image embeddings.

Building a CBIR Benchmark with TotalSegmentator and FAISS

Abstract and 1. Introduction

  1. Materials and Methods

    2.1 Vector Database and Indexing

    2.2 Feature Extractors

    2.3 Dataset and Pre-processing

    2.4 Search and Retrieval

    2.5 Re-ranking retrieval and evaluation

  2. Evaluation and 3.1 Search and Retrieval

    3.2 Re-ranking

  3. Discussion

    4.1 Dataset and 4.2 Re-ranking

    4.3 Embeddings

    4.4 Volume-based, Region-based and Localized Retrieval and 4.5 Localization-ratio

  4. Conclusion, Acknowledgement, and References

2 Materials and Methods

2.1 Vector Database and Indexing

In the context of image search a database typically constitutes the central location where all the representations of the images, a.k.a. embeddings, and their metadata including annotations are stored. A query allows the user or the system to request specific images in various ways, e.g., by inputting a reference image or a textual description. The

\ Figure 1: Overview of a retrieval system based on Khun Jush et al. [2023]: Step 1: 2D slices are extracted from the 3D volumes. Step 2: Feature extractors are used to extract the embeddings from the database slices and query volumes. Step 3: Database embeddings are indexed using HNSW or LSH indexing. Step 4: Search and slice retrieval is performed, and a hit-table is saved (the hit-table shows the occurrence of volume-ids per each query volume or region saved along with the sum of its total score). Step 5: The results from slice retrieval are aggregated to retrieve the final volume.

\ goal is to search the database for similar images that match the query. Similarly, in this study, the search process entails comparing a query image with images in the database to identify the most similar image using the cosine similarity of the embeddings. Throughout this process, we do not depend on any metadata information at any stage. Metadata-independence is an intended design choice and in contrast to widely used metadata-based image retrieval solutions that frequently lack the necessary specificity in real-world retrieval applications. In small sets, the similarity search is easy but with the growing size of the database, the complexity increases. Accuracy and speed are the key factors in search, thus, naive approaches typically fail in huge datasets.

\ Indexing in the context of content-based image search involves creating a structured system that allows for efficient storage and retrieval of images based on their visual content. A flat index is the simplest form of indexing, where no modification is made to the vectors before they are used for search. In flat indexing, the query vector is compared to every other full-size vector in the database and their distances are calculated. The nearest k of the searched spaces is then returned as the k-nearest neighbors (kNN). While this method is the most accurate, it comes at the cost of significant search time [Aumüller et al., 2020]. To improve search time, two approaches can be employed: reducing the vector size through dimensionality reduction, e.g., by reducing the number of bits representing each vector, or reducing the search scope by clustering or organizing vectors into tree structures based on similarity or distance. This results in the identification of an approximation of the true nearest neighbors, known as approximate nearest neighbor search (ANN) [Aumüller et al., 2020].

\ There are several ANN methods available. In the context of content-based volumetric medical image retrieval, Khun Jush et al. [2023] compared Locality Sensitive Hashing (LSH) Charikar [2002] and Hierarchical Navigable Small World (HNSW) Malkov and Yashunin [2018] for indexing and search. LSH hashes data points in a way that similar data points are mapped to the same buckets with higher probabilities. This allows for a more efficient search for nearest neighbors by reducing the number of candidates to be examined. HNSW [Malkov and Yashunin, 2018] indexing organizes data into a hierarchical graph structure where each layer of the hierarchy has a lower resolution. The top layer connects data points directly, but the lower layers have fewer connections. The graph structure is designed to allow for efficient navigation during the search. Compared to LSH, HNSW typically enables faster search and requires less memory Taha et al. [2024]. Based on findings in [Khun Jush et al., 2023] HSNW was chosen as the indexing method in the setting of this study due to speed advantages over LSH at a comparable recall. There are various index solutions available to store and search vectors. In this study, we used the Facebook AI Similarity Search (FAISS) package that enables fast similarity search [Johnson et al., 2019]. The indexing process involves running the feature extractors on slices of each volumetric image and storing the output embeddings per slice. The produced representations are then added to the search index which is used later on for vector-similarity-based retrieval.

2.3 Dataset and Pre-processing

We designed a CBIR benchmark relying on the publicly available TotalSegmentator (TS) dataset Wasserthal et al. [2023], version 1. This dataset comprises in total of 1204 computed tomography (CT) volumes covering 104 anatomical structure annotations (TS, V1). The anatomical regions presented in the original dataset include several fine-grained sub-classes for which we considered an aggregation to a coarser common class as a reasonable measure, e.g., all the rib classes are mapped to a single class ‘rib’. The coarse organ labels can help identify similarities and potential mismatches between neighboring anatomical regions, providing valuable insights into the proximity information of the target organ. Table 1 shows the mapping of the original TS classes to the coarse aggregated classes. For the sake of reproducibility, the query cases are sourced from the original TS test split, while the cases contained in the original TS train and validation set serve as the database for searching. The search is assessed on the retrieval rate of 29 coarse anatomical structures and 104 original TS anatomical structures.

\ The models presented in Section 2.2 are 2D models used without fine-tuning to extract the embeddings. Thus, per each 3D volume, individual 2D slices of the corresponding 3D volumes are utilized for embedding extraction. The input size for all the used models is equal to 224 × 224 pixels with image replication along the RGB channel axis. For all the ViT-based models and the ResNet50 trained on fractal images, images are normalized to the ImageNet

\ \ Figure 2: Volume-based retrieval: For a query volume Vq covering a range of anatomical regions, a volume is retrieved that should cover the same anatomical regions. The similarity search is based on all slices from the query volume.

\ \ mean and standard deviation of (.485, .456, .406) and (.229, .224, .225), respectively. For the SwinTransformer and the ResNet50 model pre-trained on the RadImageNet dataset, the images are normalized to .5 mean and .5 standard deviation based on Mei et al. [2022]. The total size of the database is 290757 embeddings, while the final query set of the test set comprises 20442 embeddings.

\

:::info Authors:

(1) Farnaz Khun Jush, Bayer AG, Berlin, Germany (farnaz.khunjush@bayer.com);

(2) Steffen Vogler, Bayer AG, Berlin, Germany (steffen.vogler@bayer.com);

(3) Tuan Truong, Bayer AG, Berlin, Germany (tuan.truong@bayer.com);

(4) Matthias Lenga, Bayer AG, Berlin, Germany (matthias.lenga@bayer.com).

:::


:::info This paper is available on arxiv under CC BY 4.0 DEED license.

:::

\

Market Opportunity
Moonveil Logo
Moonveil Price(MORE)
$0.00248
$0.00248$0.00248
+0.40%
USD
Moonveil (MORE) Live Price Chart
Disclaimer: The articles reposted on this site are sourced from public platforms and are provided for informational purposes only. They do not necessarily reflect the views of MEXC. All rights remain with the original authors. If you believe any content infringes on third-party rights, please contact service@support.mexc.com for removal. MEXC makes no guarantees regarding the accuracy, completeness, or timeliness of the content and is not responsible for any actions taken based on the information provided. The content does not constitute financial, legal, or other professional advice, nor should it be considered a recommendation or endorsement by MEXC.

You May Also Like

CME Group to launch options on XRP and SOL futures

CME Group to launch options on XRP and SOL futures

The post CME Group to launch options on XRP and SOL futures appeared on BitcoinEthereumNews.com. CME Group will offer options based on the derivative markets on Solana (SOL) and XRP. The new markets will open on October 13, after regulatory approval.  CME Group will expand its crypto products with options on the futures markets of Solana (SOL) and XRP. The futures market will start on October 13, after regulatory review and approval.  The options will allow the trading of MicroSol, XRP, and MicroXRP futures, with expiry dates available every business day, monthly, and quarterly. The new products will be added to the existing BTC and ETH options markets. ‘The launch of these options contracts builds on the significant growth and increasing liquidity we have seen across our suite of Solana and XRP futures,’ said Giovanni Vicioso, CME Group Global Head of Cryptocurrency Products. The options contracts will have two main sizes, tracking the futures contracts. The new market will be suitable for sophisticated institutional traders, as well as active individual traders. The addition of options markets singles out XRP and SOL as liquid enough to offer the potential to bet on a market direction.  The options on futures arrive a few months after the launch of SOL futures. Both SOL and XRP had peak volumes in August, though XRP activity has slowed down in September. XRP and SOL options to tap both institutions and active traders Crypto options are one of the indicators of market attitudes, with XRP and SOL receiving a new way to gauge sentiment. The contracts will be supported by the Cumberland team.  ‘As one of the biggest liquidity providers in the ecosystem, the Cumberland team is excited to support CME Group’s continued expansion of crypto offerings,’ said Roman Makarov, Head of Cumberland Options Trading at DRW. ‘The launch of options on Solana and XRP futures is the latest example of the…
Share
BitcoinEthereumNews2025/09/18 00:56
Adam Wainwright Takes The Mound Again Honor Darryl Kile

Adam Wainwright Takes The Mound Again Honor Darryl Kile

The post Adam Wainwright Takes The Mound Again Honor Darryl Kile appeared on BitcoinEthereumNews.com. Adam Wainwright of the St. Louis Cardinals in the dugout during the second inning against the Miami Marlins at Busch Stadium on July 18, 2023 in St. Louis, Missouri. (Photo by Brandon Sloter/Image Of Sport/Getty Images) Getty Images St. Louis Cardinals lifer Adam Wainwright is a pretty easygoing guy, and not unlikely to talk with you about baseball traditions and barbecue, or even share a joke. That personality came out last week during our Zoom call when I mentioned for the first time that I’m a Chicago Cubs fan. He responded to the mention of my fandom, “So far, I don’t think this interview is going very well.” Yet, Wainwright will return to Busch Stadium on September 19 on a more serious note, this time to honor another former Cardinal and friend, the late Darryl Kile. Wainwright will take the mound not as a starting pitcher, but to throw out the game’s ceremonial first pitch. Joining him on the mound will be Kile’s daughter, Sierra, as the two help launch a new program called Playing with Heart. “Darryl’s passing was a reminder that heart disease doesn’t discriminate, even against elite athletes in peak physical shape,” Wainwright said. “This program is about helping people recognize the risks, take action, and hopefully save lives.” Wainwright, who played for the St. Louis Cardinals as a starting pitcher from 2005 to 2023, aims to merge the essence of baseball tradition with a crucial message about heart health. Kile, a beloved pitcher for the Cardinals, tragically passed away in 2002 at the age of 33 as a result of early-onset heart disease. His sudden death shook the baseball world and left a lasting impact on teammates, fans, and especially his family. Now, more than two decades later, Sierra Kile is stepping forward with Wainwright to…
Share
BitcoinEthereumNews2025/09/18 02:08
XLM Price Prediction: Stellar Targets $0.26-$0.27 Range by February 2026

XLM Price Prediction: Stellar Targets $0.26-$0.27 Range by February 2026

The post XLM Price Prediction: Stellar Targets $0.26-$0.27 Range by February 2026 appeared on BitcoinEthereumNews.com. Zach Anderson Jan 14, 2026 13:31 XLM
Share
BitcoinEthereumNews2026/01/15 10:06