Copyright protects not only the original content itself but also anything which is clearly derived from the original content. That is why plagiarism doesn’t depend on whether you copied, retyped, or minorly reworded.
If you created a figure using a figure that somebody else drew as a template, then it does not matter if there are minor differences between your figure and their figure. Because your creation is closely deirved from their creation, then it is still their copyright, from both a moral and legal perspective (personally, as a scientist I find the moral perspective more compelling than the legal).
If you had simply created a figure from scratch, then it is highly unlikely that it would appear closely similar to another person’s figure – there are simply too many possible personal choices in how to diagram the same ideas, even for fairly simple ideas.
At this point, there are two ways to proceed (assuming that the rest of your content is intellectually honest):
- Create your own diagrams from scratch with significantly different content, reflecting the different ways in which you think about the area.
- Obtain rights permissions from the original publisher, and include “(figure adapted from […])” in your paper. Many publishers have an automated online method for obtaining fragment reuse rights, which can then be sent to the publisher handling your paper. If the original figure was posted in a free-reuse medium (e.g., a US or UK government report), then you may not have to do formal obtaining of rights.